LIBRARY 
 
 OF THK 
 
 UNIVERSITY OF CALIFORNIA. 
 
 . GIRT OR 
 
 ... U/TUyv-t^^ 
 94199 Class 
 
gale 'Bicentennial publications 
 STUDIES 
 
 FROM THE 
 
 CHEMICAL LABORATORY OF THE SHEFFIELD 
 SCIENTIFIC SCHOOL 
 
gale 'Bicentennial publications 
 
 With the approval of the President and Fellows 
 of Yale University, a series of volumes has been 
 prepared by a number of the Professors and In- 
 structors, to be issued in connection with the 
 Bicentennial Anniversary, as a partial indica- 
 tion of the character of the studies in which the 
 University teachers are engaged. 
 
 This series of volumes is respectfully dedicated to 
 
 0raDttatrs of tfjr Unitirrsitp 
 
STUDIES 
 
 FROM THE 
 
 CHEMICAL LABORATORY 
 
 OF THE 
 
 SHEFFIELD SCIENTIFIC SCHOOL 
 
 EDITED BY 
 
 HORACE L. WELLS 
 
 Professor of Analytical Chemistry and Metallurgy 
 
 VOLUME I. 
 
 OF THE 
 
 UNIVERSITY 
 
 OF 
 
 NEW YORK: CHARLES SCRIBNER'S SONS 
 
 LONDON: EDWARD ARNOLD 
 
 1901 
 
Copyright, 1901, 
 BY YALE UNIVERSITY 
 
 Published, October, iqoi 
 
 UNIVERSITY PRESS - JOHN WILSON 
 AND SON CAMBRIDGE, U.S.A. 
 
PREFACE 
 
 THE object of this publication is to show what the Sheffield 
 Chemical Laboratory has done and is doing in the way of 
 scientific research, and to bring together some of the more recent 
 papers in several lines of work in a form convenient for study 
 and reference. A bibliography is given which shows the work 
 of the present officials during their connection with the labora- 
 tory, while a selection of articles that have been published 
 during the past ten years, or are soon to appear, forms the main 
 part of the book. A brief historical sketch of the laboratory 
 is also presented. 
 
 H. L. W. 
 
TABLE OF CONTENTS 
 
 VOL. I. 
 
 PAGE 
 SHEFFIELD LABORATORY . 1 
 
 BIBLIOGRAPHY 4 
 
 PAPERS ON GENERAL INORGANIC CHEMISTRY: 
 
 ON A SERIES OF CAESIUM TRIHALIDES. By H. L. Wells. 
 
 INCLUDING THEIR CRYSTALLOGRAPHY. By S. L. Penfield 13 
 
 ON THE RUBIDIUM AND POTASSIUM TRIHALIDES. By H. L. 
 Wells and H. L. Wheeler. WITH THEIR CRYSTALLOGRA- 
 PHY. By S. L. Penfield 33 
 
 ON THE ALKALI-METAL PEXTAHALIDES. By H. L. Wells and 
 H. L. Wheeler. WITH THEIR CRYSTALLOGRAPHY. By 
 S. L. Penfield 48 
 
 ON SOME ALKALINE IODATES. By H. L. Wheeler. WITH 
 
 CRYSTALLOGRAPHIC NOTES. By S. L. Penfield .... 58 
 
 ON A METHOD FOR THE QUANTITATIVE DETERMINATION 
 OF CAESIUM, AND THE PREPARATION OF PURE CAESIUM 
 AND RUBIDIUM COMPOUNDS. By H. L. Wells .... 71 
 
 ON SOME PECULIAR HALIDES OF POTASSIUM AND LEAD. 
 
 ByH. L.Weils 77 
 
 ON THALLIUM TRIIODIDE AND ITS RELATION TO THE ALKALI- 
 METAL TRIIODIDES. By H. L. Wells and S. L. Penfield 84 
 
 ON SOME COMPOUNDS CONTAINING LEAD AND EXTRA 
 
 IODINE. By H. L. Wells 89 
 
 ON THE VOLUMETRIC DETERMINATION OF TITANIC ACID 
 
 AND IRON IN ORES. By H. L. Wells and W. L. Mitchell 97 
 
 ON SOME COMPOUNDS OF TRIVALENT VANADIUM. By James 
 
 Locke and Gaston H. Edwards 103 
 
 ON AN ISOMER OF POTASSIUM FfiRRiCYANiDE. By James 
 
 Locke and Gaston H. Edwards 116 
 
 ON THE FORMATION OF POTASSIUM -FERRICYANIDE 
 THROUGH THE ACTION OF ACIDS UPON THE NORMAL 
 FERRICYANIDE. By James Locke and Gaston H. Edwards 130 
 
x CONTENTS 
 
 PAPERS ON GENERAL INORGANIC CHEMISTRY 
 
 continued. PAGE 
 
 ON THE SEPARATION OF TUNGSTIC AND SILICIC ACIDS. By 
 
 H. L. Wells and F. J. Metzger 136 
 
 ON A SALT OF QUADRIVALENT ANTIMONY. By H. L. Wells 
 
 and F. J. Metzger 139 
 
 ON THE PURIFICATION OF CAESIUM MATERIAL. By H. L. 
 
 ' Wells 142 
 
 ON THE ACID NITRATES. By H. L. Wells and F. J. Metzger 146 
 INVESTIGATIONS ON DOUBLE NITRATES : 
 
 I. CESIUM DOUBLE NITRATES. By H. L. Wells and A. P. 
 
 Beardsley 151 
 
 II. CAESIUM BISMUTH NITRATE. By G. S. Jamieson . . 153 
 III. THALLOUS THALLIC NITRATE. By F. J. Metzger . . 154 
 ON CAESIUM PERIODATE AND IODATE-PERIODATE. By H. L. 
 
 Wells 155 
 
 ON THE PERIODIC SYSTEM AND THE PROPERTIES OF INOR- 
 GANIC COMPOUNDS. By James Locke 158 
 
 PAPERS ON DOUBLE HALOGEN SALTS: 
 
 ON SOME DOUBLE HALIDES OF SILVER AND THE ALKALI 
 METALS. By H. L. Wells and H. L. Wheeler. WITH 
 THEIR CRYSTALLOGRAPHY. By S. L. Penfield . . . . 207 
 
 ON THE CESIUM AND RUBIDIUM CflLORAURATES AND BROM- 
 
 AURATES. By H. L. Wells and H. L. Wheeler. WITH 
 THEIR CRYSTALLOGRAPHY. By S. L. Penfield . . . . 211 
 ON THE CAESIUM-MERCURIC HALIDES. By H. L. Wells . .218 
 ON THE CRYSTALLOGRAPHY OF THE CAESIUM-MERCURIC 
 
 HALIDES. By S. L. Penfield 236 
 
 ON THE CAESIUM- AND THE POTASSIUM-LEAD HALIDES. By 
 
 H. L. Wells 250 
 
 ON THE DOUBLE HALIDES OF TELLURIUM WITH POTASSIUM, 
 
 RUBIDIUM, AND CESIUM. By H. L. Wheeler .... 268 
 ON THE AMMONIUM-LEAD HALIDES. By H. L. Wells and 
 
 W. R. Johnston 283 
 
 ON THE RUBIDIUM-LEAD HALIDES, AND A SUMMARY OF THE 
 
 DOUBLE HALIDES OF LEAD. By H. L. Wells .... 295 
 ON THE DOUBLE HALIDES OF ARSENIC WITH CAESIUM AND 
 
 RUBIDIUM ; AND ON SOME COMPOUNDS OF ARSENIOUS 
 
 OXIDE WITH THE HALIDES OF CJESIUM, RUBIDIUM, AND 
 
 POTASSIUM. By H. L. Wheeler 300 
 
 ON SOME DOUBLE SALTS OF LEAD TETRACHLORIDE. By 
 
 H. L. Wells 313 
 
CONTENTS x i 
 
 PAPERS ON DOUBLE HALOGEN SALTS continued. 
 
 PAGE 
 ON THE DOUBLE HALIDES OF ANTIMONY WITH RUBIDIUM. 
 
 By H. L. Wheeler 320 
 
 ON THE DOUBLE CHLORIDES, BROMIDES, AND IODIDES OF 
 
 CAESIUM AND ZlNC, AND OF CAESIUM AND MAGNESIUM. 
 
 By H. L. Wells and G. F. Campbell 342 
 
 ON THE C.ESIUM-CUPRIC CHLORIDES. By H. L. Wells and 
 
 L. C. Dupee . 347 
 
 ON THE C^SIUM-CUPRIC BROMIDES. By H. L. Wells and 
 
 P. T. Walden 352 
 
 ON THE CAESIUM-CUPROUS CHLORIDES. By H. L. Wells . . 354 
 ON THE DOUBLE CHLORIDES AND BROMIDES OF CAESIUM, 
 
 RUBIDIUM, POTASSIUM, AND AMMONIUM WITH FERRIC 
 
 IRON, WITH A DESCRIPTION OF Two FERRO-FERRIC 
 
 DOUBLE BROMIDES. By P. T. Walden 357 
 
 ON THE CAESIUM-COBALT AND C^ESIUM-NlCKEL DOUBLE 
 
 CHLORIDES, BROMIDES, AND IODIDES. By G. F. Camp- 
 bell 366 
 
 ON THE DOUBLE HALIDES OF CAESIUM, RUBIDIUM, SODIUM, 
 
 AND LITHIUM WITH THALLIUM. By J. H. Pratt . . . 370 
 
 ON THE DOUBLE SALTS OF CAESIUM CHLORIDE WITH CHRO- 
 MIUM TRICHLORIDE AND WITH URANYL CHLORIDE. By 
 H. L. Wells and B. B. Boltwood 381 
 
 ON THE AMMONIUM-CUPROUS DOUBLE HALOGEN SALTS. By 
 
 H. L. Wells and E. B. Hurlburt 385 
 
 ON THE DOUBLE FLUORIDES OF CAESIUM AND ZIRCONIUM. 
 
 By H. L. Wells and H. W. Foote 390 
 
 ON CERTAIN DOUBLE HALOGEN SALTS OF CAESIUM AND 
 
 RUBIDIUM. By H. L. Wells and H. W. Foote .... 394 
 
 ON THE DOUBLE FLUORIDES OF ZIRCONIUM WITH LITHIUM, 
 SODIUM, AND THALLIUM. By H. L. Wells and H. W. 
 Foote 400 
 
 ON THE CESIUM ANTIMONIOUS FLUORIDES AND SOME OTHER 
 DOUBLE HALIDES OF ANTIMONY. By H. L. Wells and 
 F. J. Metzger 407 
 
 ON THE DOUBLE CHLORIDES OF CAESIUM AND THORIUM. 
 
 By H. L. Wells and J. M. Willis 415 
 
 ON A CJESIUM TELLURIUM FLUORIDE. By H. L. Wells and 
 
 J. M. Willis 418 
 
 GENERALIZATIONS ON DOUBLE HALOGEN SALTS. By H. L. 
 
 Wells 420 
 
 INDEX . 443 
 
SHEFFIELD LABORATORY. 
 
 THE Chemical Department of the Sheffield Scientific School 
 holds the distinction of having been the starting-point of the 
 school. The Philosophical Department of Yale College, as it 
 was at first called, formed its first class in 1847, using as a 
 laboratory the old President's House, which stood on the 
 College Campus where Farnam Hall now stands. John P. 
 Norton, Professor of Agricultural Chemistry, and Benjamin 
 Silliman, Jr., Professor of Applied Chemistry, were the first 
 instructors ; and among the earliest students were G. J. Brush, 
 W. H. Brewer, and S. W. Johnson, who have been so promi- 
 nent in the development of the school. 
 
 Yale College had taken an early prominence in chemistry 
 from the fact that Benjamin Silliman, the elder, had begun 
 his labors here in 1804. He had just returned from England, 
 where he had pursued chemical studies and had attended 
 lectures by John Dalton, the founder of the atomic theory. 
 In 1818 he founded the " American Journal of Science," which 
 has been published continuously to the present time, and is 
 one of the oldest scientific periodicals in the world. Most of 
 the publications from the Sheffield Laboratory until recent 
 times have appeared in this journal. 
 
 The chemical laboratory in the old President's House con- 
 tinued to be used for a period of thirteen years. Meanwhile 
 Professor Silliman, the younger, had severed his active con- 
 nection with it, and Professor Norton, after a highly valued 
 service of five years, had died at the early age of thirty, and 
 was succeeded by Professor John A. Porter. 
 
 Through the liberality of Joseph E. Sheffield the Chemical 
 Department, now united to an Engineering Department which 
 had existed on the College ground for a number of years, 
 was removed in 1860 to the building now known as Sheffield 
 
2 SHEFFIELD LABORATORY. 
 
 Hall. Here laboratories which were very commodious and 
 complete for their day were fitted up, and the Sheffield 
 Chemical Laboratory, gradually expanding as other depart- 
 ments were provided with new buildings, remained in this 
 place for a period of thirty-five years. 
 
 Professor Brush had been appointed to the chair of Metal- 
 lurgy in 1855, and S. W. Johnson became Professor of An- 
 alytical Chemistry in 1856, while the laboratory was still in 
 the old President's House. The entire charge of the labora- 
 tory was soon put into the hands of these two gentlemen. 
 Professor Porter, who was Mr. Sheffield's son-in-law, resigned 
 in 1864 on account of ill-health, and died two years later. 
 
 The history of the laboratory in Sheffield Hall was one of 
 steady growth and development ; a wider range of instruction 
 was gradually introduced, and from it have branched the 
 departments of Mineralogy and Physiological Chemistry. Pro- 
 fessor Brush at an early date turned his attention to mineral- 
 ogy, which he taught for many years, and in which he made 
 many important investigations, an account of which appears 
 in another volume of this series. He gradually gave up his 
 direct connection with the Chemical Department on account 
 of his duties as executive officer of the school. 
 
 Professor R. H. Chittenden began instruction in physio- 
 logical chemistry in 1875. From his efforts grew the Depart- 
 ment of Physiological Chemistry and Physiology, which for 
 a time was housed in Sheffield Hall with the chemical labo- 
 ratory, until in 1889 the acquisition of the Sheffield Mansion 
 gave it independent quarters. 
 
 Professor Johnson paid particular attention to agricultural 
 chemistry, in which he became a leading authority, and con- 
 tinued to teach this subject, as well as organic and theoreti- 
 cal chemistry, until his retirement as Professor Emeritus in 
 1895. 
 
 From 1871 to 1886 Professor O. D. Allen took charge of 
 the instruction in analytical chemistry and metallurgy. His 
 work with Professor Johnson on caesium compounds and in 
 correcting Bunsen's first determination of the atomic weight 
 
SHEFFIELD LABORATORY. 3 
 
 of caesium was very important. Professor Allen was obliged 
 to retire from the school in 1886 on account of poor health. 
 
 Professor Brewer, who was appointed to the chair of 
 Agriculture at an early date, took charge of the instruction in 
 elementary chemistry for a number of years. Since 1874 this 
 instruction has been in charge of Professor W. G. Mixter, 
 who has also carried out many investigations, particularly in 
 the lines of organic and physical chemistry. 
 
 In 1895 the Chemical Department moved from Sheffield 
 Hall into a new laboratory, which now affords the space 
 and facilities required by its growth. This building, which 
 is wholly devoted to chemistry, is one hundred and twenty- 
 nine feet long, seventy-three feet wide in front, and sixty-three 
 feet wide in the rear, and has three stories with a high 
 basement. The laboratory is finely equipped for work in 
 elementary, analytical, organic, inorganic, and physical chem- 
 istry, and contains a large chemical library. 
 
 The present chemical force consists of Professor Mixter, 
 who has been mentioned previously; the writer, who was 
 appointed as instructor in 1884, and has had charge of the 
 analytical chemistry and metallurgy since Professor Allen's 
 retirement ; Assistant Professor H. L. Wheeler, who has under 
 his care most of the investigations in organic chemistry ; Mr. 
 W. J. Comstock, who gives the greater part of the class-room 
 instruction in organic chemistry ; Assistant Professor P. T. 
 Walden, who is associated with Professor Mixter in the work 
 of instruction in elementary chemistry; Dr. James Locke, 
 instructor in inorganic chemistry; Dr. Bayard Barnes, in- 
 structor in organic chemistry ; and Dr. H. W. Foote, instruc- 
 tor in physical chemistry. 
 
BIBLIOGRAPHY. 
 
 THE list given here includes only the publications of the 
 present officials of the department, and of those who have 
 worked under their advice. This limitation has been made 
 because much of the other work has been included in the 
 volume of this series which relates to mineralogy, and because 
 some of the remaining publications were somewhat outside of 
 the domain of pure chemistry. The list is limited also to the 
 work which the authors have done in New Haven. 
 
 On Willemite and Tephroite, by W. G. Mixter. Amer. Jour. Sci. (2), xlvi, 
 pp. 230-233 (1868). 
 
 On the Estimation of Sulphur in Coal and Organic Compounds, by 
 W. G. Mixter. Ibid. (3), iv, pp. 90-95 (1872). 
 
 On Ethylidenargentamine-ethylidenammonium Nitrate, by W. G. Mix- 
 ter. Ibid., xiv, pp. 195-201 (1877). 
 
 On Amylidenamine Silver Nitrate, by W. G. Mixter. Ibid., xv, pp. 
 205-208 (1878). 
 
 On Ethylidenamine Silver Sulphate, by W. G. Mixter. Ibid., xvii, pp. 
 427-429 (1879). 
 
 On some Compounds of Aromatic Amines with Silver Nitrate and Sul- 
 phate, by W. G. Mixter. Amer. Chem. Jour., i, pp. 239-243 (1880). 
 
 On the Density of the Vapors of some Ammonium and Ammonia Com- 
 pounds, by W. G. Mixter. Ibid., ii, pp. 153-158 (1881). 
 
 Estimation of Sulphur in Illuminating Gas by Burning in Oxygen. A 
 synthesis of water for a lecture experiment, by W. G. Mixter. Ibid., 
 pp. 244-247 (1882). 
 
 On Sauer's Method of estimating Sulphur, and some Modifications, by 
 W. G. Mixter. Ibid., pp. 396-401 (1882). 
 
 On Urea from Ammonia and Carbon Dioxide, by W. G. Mixter. Ibid., 
 iv, pp. 35-38 (1882). 
 
 On some Reductions with Zinc and Ammonia, by W. G. Mixter. Ibid., 
 v, pp. 1-7 ; pp. 282-286 (1883). 
 
 On the Reduction of Benzoyl-orthonitranilide, by W. G. Mixter. Ibid., 
 vi, pp. 26-28 (1884). 
 
 Gerhardite and Artificial Basic Cupric Nitrates, by H. L. Wells and S. L. 
 Penfield. Amer. Jour. Sci., xxx, pp. 50-57 (1885). 
 
BIBLIOGRAPHY. 5 
 
 On New Acid Proprionates and Butyrates, by W. G. Mixter. Amer. 
 
 Chem. Jour., viii, pp. 343-346 (1886). 
 On Para-form-nitr-anilide, by T. B. Osborne and W. G. Mixter. Ibid., 
 
 pp. 346-347 (1886). 
 On Para-dibrom-ortho-azo-acetanilide, by C. H. Matthieson and W. G. 
 
 Mixter. Ibid., pp. 347-349 (1886). 
 On Halogen Derivatives of Oxamilide, by J. O. Dyer and W. G. Mixter. 
 
 Ibid., pp. 349-357 (1886). 
 Bismutosphserite from Willimantic and Portland, Conn., by H. L. Wells. 
 
 Amer. Jour. Sci., xxxiv, pp. 221-227 (1887). 
 Basic Lead Nitrates, by A. J. Wakeman and H. L. Wells. Amer. Chem. 
 
 Jour., ix, pp. 229-303 (1887). 
 Basic Zinc and Cadmium Nitrates, by H. L. Wells. Ibid., ix, pp. 
 
 304-308 (1887). 
 On Nitro Derivatives of Oxanilide, by W. G. Mixter and F. O. Walther. 
 
 Ibid., ix, pp. 355-361 (1887). 
 On Nitro Derivatives of Dibrom-oxanilide, by W. G. Mixter and C. P. 
 
 Willcox. Ibid., pp. 361-364 (1887). 
 Sperrylite, a New Mineral, by H. L. Wells. Amer. Jour. Sci., xxxvii, 
 
 pp. 67-73 (1889). 
 Description of the New Mineral, Beryllonite, by E. S. Dana and H. L. 
 
 Wells. Ibid., xxxvii, pp. 23-32 (1889). 
 An Elementary Text-Book of Chemistry, by W. G. Mixter. 8, viii + 459 
 
 pp. New York, 1889. [The part on Physics of Chemistry, pp. 1-45, 
 
 and Spectral Analysis, pp. 90-94, is by C. S. Hastings.] 
 On Nitro Derivatives of Oxaltoluide, by W. G. Mixter and F. Kleeberg. 
 
 Amer. Chem. Jour., xi, pp. 236-240 (1889). 
 Analyses of Several Manganesian Phosphates, by H. L. Wells. Amer. 
 
 Jour. Sci., xxxix, pp. 201-216 (1890). 
 On some Selenium and Tellurium Minerals from Honduras, by E. S. 
 
 Dana and H. L. Wells. Ibid., xl, pp. 78-82 (1890). 
 On Silver Formanilide, by W. J. Comstock and F. Kleeberg. Amer. 
 
 Chem. Jour., xii, pp. 493-502 (1890). 
 On the Composition of Pollucite and its Occurrence at Hebron, Maine, 
 
 by H. L. Wells. Amer. Jour. Sci., xli, pp. 213-220 (1891). 
 On a Self -feeding Sprengel Pump, by H. L. Wells. Ibid., xli, pp. 
 
 390-394 (1891). 
 Researches on the Isoanilides, by W. J. Comstock and H. L. Wheeler. 
 
 Amer. Chem. Jour., xiii, p. 514 (1891). 
 
 On the Preparation of the Oxygen Ethers Succinimide from its Sil- 
 ver Salt, by W. J. Comstock and H. L. Wheeler. Ibid., p. 520 
 
 (1891). 
 On some Derivatives of Aromatic Formyl Compounds, by W. J. Comstock 
 
 and R. R. Clapp. Ibid., p. 524 (189i). 
 
6 BIBLIOGRAPHY. 
 
 On a Series of Caesium Trihalides, by H. L. Wells. Including their 
 
 crystallography, by S. L. Penfield. Amer. Jour. Sci., xliii, pp. 17-32 
 
 (1892). 
 On the Rubidium and Potassium Trihalides, by H. L. Wells and H. L. 
 
 Wheeler. With their crystallography, by S. L. Penfield. Ibid., xliii, 
 
 pp. 476-487 (1892). 
 On the Alkali-metal Pentahalides, by H. L. Wells and H. L. Wheeler. 
 
 With their crystallography, by S. L. Penfield. Ibid., xliv, pp. 42-49 
 
 (1892). 
 On Herderite from Hebron, Maine, by H. L. Wells and S. L. Penfield. 
 
 Ibid., xliv, pp. 114-116 (1892). 
 On some Double Halides of Silver and the Alkali-metals, by H. L. Wells 
 
 and H. L. Wheeler. W^ith their crystallography, by S. L. Penfield. 
 
 Ibid., xliv, pp. 155-157 (1892). 
 On the Caesium and Rubidium Chloramates and Bromamates, by H. L. 
 
 Wells and H. L. Wheeler. W r ith their crystallography, by S. L. 
 
 Penfield. Ibid., xliv, pp. 157-162 (1892). 
 On some Alkaline lodates, by H. L. W'heeler. With crystallographic 
 
 Notes, by S. L. Penfield. Ibid., xliv, pp. 124-132 (1892). 
 On the Caesium-Mercuric Halides, by H. L. Wells. Ibid., xliv, pp. 
 
 221-236 (1892). 
 On the Caesium-Lead and the Potassium-Lead Halides, by H. L. Wells. 
 
 Ibid., xlv, pp. 121-134 (1893). 
 On the Double Halides of Tellurium with Potassium, Rubidium, and 
 
 Caesium, by H. L. Wheeler. Ibid., xlv, pp. 267-269 (1893). 
 On the Ammonium-Lead Halides, by H. L. Wells and W. R. Johnston. 
 
 Ibid., xlvi, pp. 25-34 (1893). 
 On the Rubidium-Lead Halides, and a Summary of the Double Halides 
 
 of Lead, by H. L. Wells. Ibid., xlvi, pp. 34-38 (1893). 
 On the Double Halides of Arsenic with Caesium and Rubidium ; and on 
 
 some Compounds of Arsenious Oxide with the Halides of Caesium, 
 
 Rubidium, and Potassium, by H. L. Wheeler. Ibid., xlvi, pp. 88-98 
 
 (1893). 
 On the Deportment of Charcoal with the Halogens, Nitrogen, Sulphur, 
 
 and Oxygen, by W. G. Mixter. Ibid., pp. 363-369 (1893). 
 On some Double Salts of Lead Tetrachloride, by H. L. Wells. Ibid., 
 
 xlvi, pp. 180-186 (1893). 
 On a Method for the Quantitative Determination of Caasium, and the 
 
 Preparation of Pure Caesium and Rubidium Compounds, by H. L. 
 
 Wells. Ibid., xlvi, pp. 186-190 (1893). 
 On some Peculiar Halides of Potassium and Lead, by H. L. Wells. Ibid., 
 
 xlvi, pp. 190-195 (1893). 
 On the Double Halides of Antimony with Rubidium, by H. L. Wheeler. 
 
 Ibid., xlvi, pp. 269-279 (1893). 
 
BIBLIOGRAPHY. 1 
 
 On the Double Chlorides, Bromides, and Iodides of Caesium and Cad- 
 mium, by II. L. Wells and P. T. Walden. Amer. Jour. Sci., xlvi, 
 
 pp, 425-431 (1893). 
 On the Double Chlorides, Bromides, and Iodides of Caesium and Zinc, 
 
 and of Caesium and Magnesium, by H. L. Wells and G. F. Campbell. 
 
 Ibid., pp. 431-434 (1893). 
 On the Caesium-Cupric Chlorides, by H. L. Wells and L. C. Dupee. Ibid., 
 
 xlvii, pp. 91-93 (1894). 
 On the Csesium-Cupric Bromides, by H. L. Wells and P. T. Walden. 
 
 Ibid., xlvii, pp. 94-96 (1894). 
 On the Caesium-Cuprous Chlorides, by H. L. Wells. Ibid., xlvii, pp. 
 
 96-98 (1894). 
 On Thallium Triiodide and its Relation to the Alkali-Metal Triiodides, 
 
 by H. L. Wells and S. L. Penfield. Ibid., xlvii, pp. 463-466 (1894). 
 On the Occurrence of Leadhillite in Missouri and its Chemical Composi- 
 tion, by L. V. Pirsson and H. L. Wells. Ibid., xlviii, pp. 219-226 
 
 (1894). 
 On the Double Chlorides and Bromides of Caesium, Rubidium, Potassium, 
 
 and Ammonium with Ferric Iron, with a Description of two Ferro- 
 
 ferric double bromides, by P. T. Walden. Ibid., xlviii, pp. 283-290 
 
 (1894). 
 On the Caesium-Cobalt and Caesium-Nickel Double Chlorides, Bromides, 
 
 and Iodides, by G. F. Campbell. Ibid., xlviii, pp. 418-420 (1894). 
 On some Azo and Azimido Compounds, by W. G. Mixter. Amer. Chem. 
 
 Jour., xvii, pp. 449-453 (1895). 
 On the Double Halides of Caesium, Rubidium, Sodium, and Lithium 
 
 with Thallium, by J. II. Pratt. Amer. Jour. Sci., xlix, pp. 397-404 
 
 (1895). 
 On some Compounds containing Lead and Extra Iodine, by H. L. Wells. 
 
 Ibid., 1, pp. 21-26 (1895). 
 On the Double Salts of Caesium Chloride with Chromium Trichloride, 
 
 and with Uranyl Chloride, by H. L. Wells and B. B. Boltwood. Ibid., 
 
 1, pp. 249-252 (1895). 
 On the Ammonium-Cuprous Double Halogen Salts, by H. L. Wells and 
 
 E. B. Hurlburt. Ibid., 1, pp. 390-393 (1895). 
 On the Volumetric Determination of Titanic Acid and Iron in Ores, by 
 
 H. L. Wells and W. L. Mitchell. Jour. Amer. Chem. Soc., xvii, pp. 
 
 878-883 (1895). 
 On the Double Fluorides of Caesium and Zirconium, by H. L. Wells and 
 
 H. W. Foote. Amer. Jour. Sci. (4), i, pp. 18-20 (1896). 
 On Halogen Addition-Products of the Anilides, by H. L. Wheeler and 
 
 P. T. Walden. Amer. Chem. Jour., xviii, p. 85 (1896). 
 The Action of Acid Chlorides on the Silver Salts of the Anilides, by 
 
 H. L. Wheeler and B. B. Boltwood. Ibid., xviii, p. 381 (1896). 
 
8 BIBLIOGRAPHY. 
 
 On some Mercury Salts of the Anilides, by H. L. Wheeler and B. W. 
 
 McFarland. Amer. Chem. Jour., xviii, p. 540 (1896). 
 On the Use of Antimony Trichloride in the Synthesis of Aromatic Ketones 
 
 by W. J. Comstock. Ibid., pp. 547-552 (1896). 
 On Diacid Anilides, by H. L. Wheeler. Ibid., xviii, p. 695 (1896). 
 On the Action of Acid Chlorides on the Imido Esters and Isoanilides and 
 
 on the Structure of the Silver Salts of the Anilides, by H. L. Wheeler 
 
 and P. T. Walden. Ibid., xix, p. 129 (1897). 
 On the Action of Chlorcarbonic Ethyl Ester on Formanilide, by H. L. 
 
 Wheeler and H. F. Metcalf. Ibid., xix, p. 217 (1897). 
 On the Preparation of Metabrombenzoic Acid and of Metabromnitro- 
 
 benzene, by H. L. Wheeler and B. W. McFarland. Ibid., xix, p. 363 
 
 (1897). 
 On the Non- Existence of Four Methenylphenylparatolyl Amidines, by 
 
 H. L. Wheeler. Ibid., xix, p. 365 (1897). 
 On the Molecular Rearrangement of the Oxines by Means of Certain 
 
 Metallic Salts, by W. J. Comstock. Ibid., xix, pp. 485-492 (1897). 
 On Electrosynthesis, by W. G. Mixter. Amer. Jour. Sci., iv, pp. 51-62 
 
 (1897). 
 On Halogen Addition Products of the Anilides, by H. L. Wheeler, Bayard 
 
 Barnes, and J. H. Pratt. Amer. Chem. Jour., xix, p. 672 (1897). 
 On Diacyl Anilides, by H. L. Wheeler, T. E. Smith, and C. H. Warren. 
 
 Ibid., xix, p. 757 (1897). 
 Fresenius's Manual of Qualitative Chemical Analysis. Translated by 
 
 H. L. Wells. 8vo, pp. xvii, 748 (Xew York, 1897). 
 On Certain Double Halogen Salts of Caesium and Rubidium, by H. L. 
 
 Wells and H. W. Foote. Amer. Jour. Sci., iii, pp. 461-465 (1897). 
 On the Double Fluorides of Zirconium with Lithium, Sodium, and Thal- 
 lium, by H. L. Wells and H. W. Foote. Ibid., iii, pp. 466-471 (1897). 
 On Acylimidoesters, by H. L. Wheeler, P. T. Walden, and H. F. Metcalf. 
 
 Amer. Chem. Jour., xx, p. 64 (1898). 
 Note on Double Salts of the Anilides with Cuprous Chloride and Cuprous 
 
 Bromide, by W. J. Comstock. Ibid., pp. 77-79 (1898). 
 On some Bromine Derivatives of 2-3-Dimethylbutane, by H. L. 
 
 Wheeler. Ibid., xx, 148 (1898). 
 On the Silver Salt of 4-Nitro-2-Arainobenzoic Acid and its Behavior with 
 
 Alkyl and Acyl Halides, by H. L. Wheeler and Bayard Barnes. 
 
 Ibid., xx, 217 (1898). 
 On the Action of Hydrogen Sulphide upon Vanadates, by James Locke. 
 
 Ibid., xx, pp. 373-376 (1898). 
 Researches on the Cycloamidines : Pyrimidine Derivatives, by H. L. 
 
 Wheeler. Ibid., xx, 481 (1898). 
 A Laboratory Guide in Qualitative Chemical Analysis, by H. L. Wells, 
 
 8vo, pp. vi, 189 (New York, 1898). 
 
BIBLIOGRAPHY. 9 
 
 Researches on the Cycloamides : a-Ketobenzmorpholine and a-Benzpara- 
 
 oxazine Derivatives, by H. L. Wheeler and Bayard Barnes. Amer. 
 
 Chem. Jour., xx, 555 (1898). 
 The Action of Amines on Acylimido Esters : Acyl Amidines, by H. L. 
 
 Wheeler and P. T. Walden. Ibid., xx, 568 (1898). 
 On the Periodic System and the Properties of Inorganic Compounds, by 
 
 James Locke. Ibid., xx, pp. 581-592 (1898). 
 The Action of Sulphur upon Metallic Sodium, by James Locke and 
 
 Alfred Austell. Ibid., xx, pp. 592-594 (1898). 
 On some Compounds of Trivalent Vanadium, by James Locke and G. 
 
 H. Edwards. Ibid., xx, 594-606 (1898). 
 On Electrosynthesis (Second Paper), by W. G. Mixter. Amer. Jour. Sci., 
 
 vi, pp. 217-224 (1898). 
 On the Non-Existence of Four Methenylphenylparatolylamidines, by 
 
 H. L. Wheeler and T. B. Johnson. Amer. Chem. Jour., xx, 853 (1898). 
 On the Rearrangement of Imidoesters, by H. L. Wheeler and T. B. 
 
 Johnson. Ibid., xxi, 185 (1899). 
 On an Isoiner of Potassium Ferricyanide, by James Locke and G. H. 
 
 Edwards. Ibid., xxi, pp. 193-206 (1899). 
 On the Formation of Potassium /3-Ferricyanide through the Action of 
 
 Acids upon the Normal Ferricyanide, by James Locke and G. H. 
 
 Edwards. Ibid., xxi, pp. 413-418 (1899). 
 On some Experiments with Endothermic Gases, by W. G. Mixter. Amer. 
 
 Jour. Sci., vii, pp. 323-327 (1899). 
 On a Hypothesis to Explain the Partial Non-Explosive Combination of 
 
 Explosive Gases and Gaseous Mixtures, by W. G. Mixter. Ibid., pp. 
 
 327-334 (1899). 
 On the Rearrangement of the Thioncarbamic Esters, by H. L. Wheeler 
 
 and Bayard Barnes. Amer. Chem. Jour., xxi, 141 (1899). 
 Researches on Substitution : the Action of Bromine on Metachlor-, 
 
 metabrom-, and metaiodanilines, by H. L. Wheeler and William 
 
 Valentine. Ibid., xxii, 266 (1899). 
 On the Products of Explosion of Acetylene, by W. G. Mixter. Amer. 
 
 Jour. Sci., ix, pp. 1-8 (1900). 
 On the Rearrangement of Imidoesters (Second Paper), by H. L. Wheeler. 
 
 Arner. Chem. Jour., xxiii, 135 (1900). 
 Researches on the Sodium Salts of the Amides, by H. L. Wheeler. Ibid., 
 
 xxiii, 453 (1900). 
 On the Molecular Rearrangement of the Thioncarbamic, Thioncar- 
 
 banilic, and Thioncarbazinic Esters : /3-Alkyl-a-/t-Diketotetrahydrothi- 
 
 azoles, by H. L. Wheeler and Bayard Barnes. Ibid., xxiv, 60 (1900). 
 On Ureaimido Esters, Thioureaimido Esters, Acylthioureaimido Esters, 
 
 and Ureaamidines, by H. L. Wheeler and W. Murray Sanders. 
 
 Jour. Amer. Chem. Soc., xxii, 365 (1900). 
 
10 BIBLIOGRAPHY. 
 
 On the Behavior of Acylthioncarbamic Esters towards Alkyl Iodides 
 
 and Amines : Benzoylimidothiocarbonic Esters, Acyclic Benzoylpseu- 
 
 doureas and Benzoylureas, by H. L. Wheeler and T. B. Johnson. 
 
 Am. Chem. Jour., xxiv, 189 (1900). 
 On the Products of the Explosion of Acetylene, and of Mixtures of 
 
 Acetylene and Nitrogen (Second Paper), by W. G. Mixter. Am. 
 
 Jour. Sci., x, pp. 299-309 (1900). 
 On the Molecular Rearrangement of Disubstituted Thioncarbamic Ester : 
 
 Phenylimidocarbonic Acid Derivatives and Thiosemicarbazidic Esters, 
 
 by H. L. Wheeler and Guy K. Dustin. Ibid., xxiv, 424 (1900). 
 On the Action of Alkylthiocyanates and Alkylisothiocyanates with Thiol- 
 
 Acids, by H. L. Wheeler and H. F. Merriam * 
 On Acetyl and Benzoylmidodithiocarbonic Esters, by H. L. Wheeler 
 
 and T. B. Johnson.* 
 
 Kesearches on Thiocyanates and Isothiocyanates, by H. L. Wheeler.* 
 On some Acetyl and Benzoylpseudothioureas, by H. L. Wheeler and 
 
 T. B. Johnson.* 
 On some Addition Reactions of Thio Acids, by H. L. Wheeler and 
 
 Bayard Barnes.* 
 On the Caesium Antimonious Fluorides and some other Double Halides 
 
 of Antimony, by H. L. Wells and F. J. Metzger.* 
 On the Separation of Tungstic and Silicic Acids, by H. L. Wells and 
 
 F. J. Metzger.* 
 
 On a Salt of Quadrivalent Antimony, by H. L. Wells and F. J. Metzger.* 
 On the Purification of Caesium Material, by H. L. Wells.* 
 On the Acid Nitrates, by H. L. Wells and F. J. Metzger.* 
 Investigations on Double Nitrates : I. Caesium Double Nitrates, by H. L. 
 
 Wells and A. P. Beardsley ; II. Caesium Bismuth Nitrate, by G. J. 
 
 Jamieson ; III. Thallous Thallic Nitrate, by F. J. Metzger * 
 On Caesium Periodate and lodate-Periodate, by H. L. Wells.* 
 On the Double Chlorides of Caesium and Thori^un, by H. L. Wells and 
 
 J. M. Willis.* 
 
 On a Csesium Tellurium Fluoride, by H. L. Wells and J. M. Willis.* 
 On the Periodic System and the Properties of Inorganic Compounds : II. 
 
 Gradations in the Properties of Alums, by James Locke.* 
 On the Periodic System and the Properties of Inorganic Compounds: 
 
 III. The Solubilities of Alums as a Function of Two Variables, by 
 
 James Locke.* 
 Generalizations on Double Halogen Salts, by H. L. Wells.* 
 
 * To be published. 
 
PAPERS ON 
 GENERAL INORGANIC CHEMISTRY 
 
ON A SERIES OF OESIUM TRIHALIDES.* 
 
 BY H. L. WELLS. 
 
 INCLUDING THEIR CRYSTALLOGRAPHY. 
 BY S. L. PENFIELD. 
 
 IN the course of some experiments with caesium compounds, 
 bromine was added to a concentrated solution of caesium chlo- 
 ride with an astonishing result. There was instantly formed 
 a bright yellow precipitate, so dense as to nearly solidify the 
 liquid. The substance readily dissolved on warming the 
 liquid, and, on cooling it, large crystals of a yellowish-red 
 color were formed which were found to be CsClBr 2 . 
 
 In view of the fact that KI 3 was already known, f this dis- 
 covery made it probable that a series of caesium trihalides 
 could be obtained. An attempt was accordingly made to 
 prepare each of the following possible members of such a 
 series containing chlorine, bromine, and iodine. 
 
 1. CsI 3 6. CsCl 2 I 
 
 2. CsBrI 2 7. CsBr 3 
 
 3. CsBr 2 I 8. CsClBr 2 
 
 4. [CsClI, 9. CsCl 2 Br 
 
 5. CsClBrI 10. [CsCl,] 
 
 As a result, all the members of the series except the two 
 enclosed in brackets were isolated. 
 
 These eight trihalides are easily made, being much less 
 soluble than the normal halides. They crystallize beautifully, 
 have remarkably brilliant colors, and some of them possess an 
 unexpected degree of stability. 
 
 * Amer. Jour. Sci., xliii, January, 1892. 
 
 t Jorgensen, J. pr. Ch., II, ii, 357 ; Johnson, J. Chem. Soc., 1877, 249. 
 
14 ON A SERIES OF 
 
 Method of Preparation. 
 
 Each of these compounds can be made by dissolving, with 
 the aid of heat, the appropriate normal csesium halide and 
 the halogen or halogens indicated by the formula in the proper 
 amount of water, or, in the single case of CsBrI 2 , in weak 
 alcohol, and cooling to crystallization. The caesium salt used 
 in making the mixed trihalides is preferably the one which 
 is not decomposed by the halogen or halogens added. In 
 most cases the presence of an excess of the normal halide is 
 desirable in order that the halogens, especially iodine, may 
 readily dissolve and not separate again on cooling, but 
 the same result may also be obtained by the use of weak 
 alcohol. Details of preparation will be given for each body 
 separately. 
 
 Color. 
 
 In the following list the compounds are arranged in order, 
 from the darkest to the lightest. The colors given, unless 
 otherwise specified, are for crystals of considerable size, for 
 when the bodies are obtained as precipitates, or when the 
 crystals are pulverized, they are lighter in color. 
 
 CsI 3 Brilliant black, nearly opaque ; powder brown. 
 
 CsBrI 2 Dark reddish-brown; thin crystals transmit 
 
 deep red light ; powder dark red. 
 CsBr 2 I Deep cherry-red. 
 
 g I Yellowish-red, each having a somewhat 
 C GIB j yellower tint than the one preceding it. 
 
 ( Orthorhombic variety, deep orange. 
 82 ( Rhombohedral variety, pale orange. 
 CsCl 2 Br Bright yellow. 
 
 Stability on Exposure. 
 
 The five bodies containing iodine are much more stable 
 than the others, and will bear long exposure to the air at 
 ordinary temperatures with very slight superficial change. 
 This exposure in some cases may be continued for a week or 
 
CESIUM TRIHALIDES. 15 
 
 more in warm weather without producing any marked alter- 
 ation of color, but they constantly give off a slight odor and 
 finally begin to whiten. The three compounds containing 
 no iodine usually become white in a few hours on exposure, 
 but even these can be preserved indefinitely in tightly corked 
 tubes. Experiments showed that CsBrI 2 whitened more 
 rapidly than CsBr 2 I, also that CsClBr 2 decomposed more 
 rapidly than CsCl 2 Br. This indicates that their stability 
 does not entirely depend upon the volatility of the halogens 
 contained in them, a point which has a bearing on the con- 
 stitution of this group of bodies, and which will be considered 
 subsequently. 
 
 Behavior when Heated. 
 
 The following table shows the temperatures of complete 
 decomposition as determined by the change of color to white. 
 They are only approximate, since they represent gradual 
 changes which vary somewhat with the rapidity of heating. 
 The melting-points are also given. In open tubes these are 
 usually sharp, but in sealed tubes often very gradual. 
 
 Melts Melts Becomes white 
 
 in open tube. in sealed tube. in open tube. 
 
 (uncorr.) (uncorr.) (approximate.) 
 
 CsI 8 210 201-208 330 
 
 CsBr 2 I 246 243-248 320 
 
 CsClBrI 238 225-235 290 
 
 CsCl 2 I 238 225 -230 290 
 
 CsBrI 2 208 155-190 260 
 
 CsBr 8 whitens 180 160 
 
 CsCl 2 Br whitens 205 150 
 
 CsClBr, whitens 191 150 
 
 Behavior with Solvents. 
 
 All these bodies except CsBrI 2 , which is almost completely 
 decomposed by water, can be recrystallized by treating with 
 warm water and cooling the solution. There is usually some 
 decomposition during this operation, accompanied by the 
 separation of iodine or the volatilization of this or the other 
 halogens. 
 
16 ON A SERIES OF 
 
 All the trihalides containing iodine can be dissolved in 
 alcohol and recrystallized from it. There is usually a slight 
 deposition of normal halide at the same time, which can be 
 avoided by adding a little water to the alcohol. CsI 3 is much 
 more soluble in alcohol than in water. The other iodine com- 
 pounds, with the exception of CsBrI 2 , which decomposes with 
 water, are apparently more soluble in water. Those bodies 
 containing no iodine are all decomposed by alcohol, leaving 
 a white residue. Mixtures of alcohol and water are good sol- 
 vents for all the trihalides. 
 
 Ether has no immediate action on the more stable com- 
 pounds, CsI 8 , CsBr 2 I, CsClBrI, and CsClJ, but it decomposes 
 all the others with separation of normal halides. When 
 CsBrI 2 is thus decomposed, pure CsBr is left. 
 
 Crystallograp Jiy. 
 
 The crystallization of the caesium trihalides is orthorhom- 
 bic. The salts form an isomorphous group, the chief features 
 of which will first be given, followed by a brief description 
 of the different individuals. 
 
 The forms which have been observed are : 
 
 a, 100, i-1 g, 012, -T 
 
 b, 010, i-i d, Oil, 1-T 
 
 c, 001, /, 021, 2-i 
 m, 110, / e, 102, \-1 
 
 Of these m, c?, and e are the most prominent and usually 
 determine the habit of the crystals. Either m or d usually 
 predominates to such an extent that the crystals are prismatic 
 in the direction of the vertical or the brachy-axes. The dome 
 / is very common, but is usually too small to give a character- 
 istic habit, and is therefore omitted from most of the figures. 
 The face g was observed only on CsI 8 . The pinacoids are 
 variable in their development, but commonly one, and fre- 
 quently all three, can be found on a single crystal. Pyram- 
 idal faces are practically wanting. In the examination of 
 a great many crystals, but one was found (of CsBr 2 I) on 
 
CJ2SIUM TR1HALIDES. 
 
 17 
 
 which a single pyramidal face occurred ; this replaced the edge 
 between m and d, and had the symbol 132, f 3. The cleav- 
 age is perfect, parallel to <?. 
 
 An idea of the similarity of the different salts may be 
 obtained from the following table, in which the axial ratios * 
 and three of the prominent angles are given. The angles 
 which were chosen as fundamental are marked by an asterisk, 
 the others are calculated, and in all cases the measurements 
 showed close agreement. 
 
 
 i 
 
 
 
 a:b: 
 
 c 
 
 
 f CsI 3 0.6824:1 
 
 1.1051 
 
 Series 
 with iodine 
 
 CsBrI 2 0.6916 : 1 
 ^ CsBr 2 I 0.7203 : 1 
 CsClBrI 0.7230 : 1 
 
 1.1419 
 1.1667 
 1.1760 
 
 
 I CsCl 2 I 0.7373 : 1 
 
 1.1920 
 
 Series 
 without iodine 
 
 , CsBr 8 0.6873 : 1 
 ] CsClBr 2 0.699 : 1 
 <CsCl 2 Br 0.7186:1 
 
 1.0581 
 1.1237 
 
 
 nt A m, 110 A HO 
 
 dAd, Oil A Oil. 
 
 e A e, 102 A T02. 
 
 CsI 3 
 CsBrI 2 
 CsBrJ 
 
 *68 37' 
 *69 20' 
 *71 32' 
 
 *95 43' 
 97 34' 
 
 *98 48' 
 
 78 0' 
 *79 5' 
 78 0' 
 
 CsClBrI 
 
 *71 44' 
 
 *99 15' 
 
 78 15' 
 
 CsCl 2 I 
 CsBr 3 
 
 CsClBr 2 
 CsCl 2 Br 
 
 72 48' 
 *69 0' 
 *69 56' 
 *71 24' 
 
 *100 1' 
 *93 14' 
 
 *96 40' 
 
 *77 54' 
 *75 10' 
 
 76 0' 
 
 CsI 8 and CsBr s are almost identical in axial ratios, but it 
 is surprising that the chemically intermediate compounds, 
 CsBrI 2 and CsBr 2 I, are not crystallographically intermediate. 
 
 A very remarkable relation exists between the first five 
 compounds in the table, all of which contain iodine and are 
 
 * The ratios are given in two ways : I, with 6 as unity, as is customary ; 
 and II, with a as unity, in order to show more clearly the relation between d 
 and c. 
 
 2, 
 
18 ON A SERIES OF 
 
 arranged in the order of their molecular weights. The ratio 
 of two axes remains nearly constant throughout, while the 
 third varies.* Exactly the same relation exists between the 
 three compounds containing no iodine, but in this series 
 the ratio between the two constant axes varies slightly from 
 the corresponding ratio in the iodine compounds. 
 
 If an arrangement according to molecular weights is made 
 with all the compounds containing bromine, or in like manner 
 with all those containing chlorine, a symmetrical series of axial 
 ratios is not formed. This leads to the conclusion that the 
 two series given in the table have a special significance, and 
 that iodine, with the highest atomic weight, plays an important 
 part in the constitution of the first, while bromine acts in the 
 same way in the second. Since several of the compounds con- 
 tain only a single halogen atom of highest atomic weight, it 
 follows that a single atom throughout exerts an influence on 
 the symmetry of the series. This peculiar part played by an 
 iodine or bromine atom may be explained by supposing it to be 
 closely united either with the caesium or with one of the other 
 halogen atoms. It seems probable from these considerations 
 that the three halogen atoms in these compounds do not have 
 similar positions in the molecule, and consequently that the 
 trihalides are not compounds of trivalent caesium but have 
 some other structure. 
 
 Csl s . Of this salt, crystals from both aqueous and alco- 
 holic solutions were examined. On the former the forms , e, 
 w, d, and / were observed. The habit was different from any- 
 thing else in the series, being needle-like, with a and m in the 
 prismatic zone, terminated by d, Fig. 1, or by <?, c?, and/, Fig. 
 2. The crystals did not give very satisfactory reflections, but 
 the best measurements, from a number of selected crystals, 
 agreed closely with those given in the table. The crystals 
 examined were 20-30 mm. in length and seldom over 2 mm. 
 in diameter. On the crystals from alcohol, the forms #, 5, c, 
 m, g, d,f, and e were observed. The habit is shown in Fig. 3. 
 
 * Several series of organic compounds with analogous axial relations have 
 been observed by Groth (Berichte, iii, 449) and by Hintze (Berichte, vi, 593). 
 
CAESIUM TRIHALIDES. 
 
 19 
 
 There was a tendency in the crystals to arrange themselves in 
 parallel position, forming plates showing large a faces, but the 
 
 772 
 
 m 
 
 m 
 
 separate individuals were small, scarcely over 3 mm. in greatest 
 diameter. The faces gave excellent reflections. The crystals 
 are black, transmit a brownish-red light only on the thinnest 
 edges, and are too opaque for optical examination. 
 
 CsBrI z . On two separate samples of this salt the forms a, 
 >, c, m, d, /, and e were observed. The crystals are thin tables 
 somewhat lengthened in the direction of 
 the brachy-axis, Fig. 4. On examining 
 the general table, it will be seen that this 
 is the only one of the first five salts in 
 which the angle e A e varites considerably 
 from 78. Here the variation amounts 
 to a little over one degree, but in all 
 probability this is not to be accounted 
 for by imperfections in the crystals or 
 inaccuracy in the observation, for from 
 two different crops of crystals good reflections and almost 
 identical measurements were obtained. The crystals were 
 only a fraction of a millimetre in thickness, and not over 10 
 mm. long in the direction of the brachy-axis. With the polar- 
 izing microscope the tables show a decided pleochroism. For 
 
20 
 
 ON A SERIES OF 
 
 rays vibrating parallel to the c axis the color is dark brown, 
 almost opaque, while for vibrations parallel to & it is a rich 
 reddish brown. A similar though less marked pleochroism 
 was observed in the remaining salts of the series, but owing 
 to the inability to obtain orientated pinacoid sections, it could 
 not be studied satisfactorily. In convergent polarized light 
 the phenomena were not very distinct, but with the tables 
 of CsBrI 2 apparently an obtuse bisectrix could be seen, the 
 optical axis being in the macro-pinacoid &. 
 
 CsBr 2 I. On this salt the forms a, <?, m, d, /, and e were 
 observed. The habit is shown in Fig. 5. The crystals were 
 brilliant and gave excellent reflections. Those submitted for 
 measurement were about 3 mm. in greatest diameter. 
 
 Os OlBrl. On this salt the forms , c, m, d, /, and e were 
 observed. The habit is like Fig. 5, but much longer, or pris- 
 matic, in the direction of the brachy- 
 axis. The crystals were about 
 2 mm. in diameter and 10 in length. 
 CsCl z I. This compound is di- 
 morphous. 
 
 On the orthorhombic modification 
 the forms a, <?, TW, c?,/, and e were 
 observed, but m and / are usually 
 wanting. The habit is shown in 
 Fig. 6. The crystals were about 
 2 mm. in diameter. 
 
 The hexagonal, rhombohedral variety occurred in curious 
 saddle-shaped scales, with bright crystal faces only along 
 
CAESIUM TRIIIALIDES. 21 
 
 the edges. The forms which were observed are r (1, lOll), 
 / (-2, 0221), and a (i-2, 1120), Fig. 7. Only that portion is 
 perfect which is included between the irregular dotted lines, 
 the upper and lower angles of the rhombohedrons being 
 truncated by irregular warped surfaces. In their growth the 
 individuals of a whole series of crystals, with similar orienta- 
 tion, are piled upon one another in the direction of the vertical 
 axis. The scales are about 6 mm. in diameter. 
 
 The measurement which was taken as fundamental is : 
 
 r A r, over a, 10T1 A 01TT = 99 48' 
 
 giving for the length of the vertical axis, c = 0.96363. The 
 following measurements were also made : 
 
 a A a = 60 0', 60 1', 59 59' Calculated 60 0' 
 a A/, I2TO A 022l = 37 45' " 37 49' 
 
 The remaining compounds containing no iodine were much 
 more unstable than those previously described. By making 
 the measurements in a cold room very satisfactory results 
 were obtained, the crystals retaining enough lustre to give 
 good reflections, even after they had suffered considerable 
 decomposition. 
 
 OsBr 8 . The forms observed on this salt are 5, m, and d. 
 The habit is short prismatic, with either m or d predominating, 
 while b is usually wanting. Single crystals are sometimes 
 10 mm. in length, but groups of small crystals are more apt to 
 occur. 
 
 Cs ClBrz. The only forms observed on this salt are m and 
 c. The habit is short and stout prismatic. Single crystals are 
 sometimes 15 mm. in length. 
 
 CsChBr. The forms observed on this salt are <?, TTZ, and d. 
 It crystallized in stout prisms, over 10 mm. long, terminated 
 like Fig. 1. 
 
 Method of Analysis. 
 
 The samples were prepared for analysis by pressing on 
 paper. The drying was not always very good, both on 
 
22 ON A SERIES OF 
 
 account of the haste sometimes necessary to avoid too much 
 decomposition, and on account of the great tendency of the 
 crystals to contain cavities filled with liquid. 
 
 Caesium was invariably determined by weighing the nor- 
 mal halide produced by heating. In some cases where the 
 resulting normal halide was slightly contaminated by a 
 higher halogen, this was replaced by the proper one before 
 weighing. 
 
 Where two halogens were present, they were determined in 
 the usual way by weighing their silver salts and determining 
 the loss in weight of these when heated in chlorine. In the 
 cases where all three halogens were present, use was made of 
 the extremely satisfactory method described by Gooch and 
 Ensign.* 
 
 Csl s . 
 
 This can be made by dissolving about one-fourth the theo- 
 retical amount of iodine in a solution of one part of cassium 
 iodide in ten parts of water. It generally gives a crop of 
 brilliant, slender crystals. If a larger proportion of iodine is 
 used, the substance generally separates in the form of crystal- 
 line plates without distinct faces. They are possibly a dimor- 
 phous form of the substance. If weak alcohol is used as a 
 solvent instead of water, the theoretical amount of iodine can 
 be taken and a well-crystallized product is obtained. 
 
 The following numbers show the composition : 
 
 Found. Calculated for 
 
 Slender Crystals. Plates. CsI 3 . 
 
 Csesium 25.41 23.71 25.88 
 
 Iodine 72.67 . . . 74.12 
 
 When iodine is being dissolved in a warm aqueous solution 
 of Csl, or when an attempt is made to dissolve Csl s in warm 
 water, a heavy black liquid is formed at about 73 which 
 solidifies on cooling to a crystalline mass. It is much richer 
 in iodine than CsI 8 and probably contains a higher polyiodide. 
 Analyses of the substance gave varying results, and although 
 
 * Amer. Jour. Sci., III, xl, 145, 1890. 
 
CAESIUM TRIHALIDES. 23 
 
 most of these approached the composition CsI 6 , it is still un- 
 certain what body this is. The low melting-point of the 
 substance is remarkable, since CsI 3 melts at 210 and iodine 
 at 114. 
 
 To find the solubility of CsI 3 in an aqueous solution of Csl, 
 the mother-liquor from a crop of crystals deposited at about 
 20 was analyzed. It gave : 
 
 Csl 16.99 per cent. 
 
 I (free) 0.416 per cent or Csl s 0.842 per cent. 
 
 The specific gravity of this mother-liquor was 1.154, hence 
 1 c. c. contained 0.0097gCsI 3 . The body is so insoluble that, 
 were we to use Csl in the place of KI in making a volumetric 
 iodine solution, we could only obtain, at ordinary temperatures, 
 a solution which would be about ^ normal. 
 
 OsBrI 2 . 
 
 Iodine dissolves in considerable quantity in a hot aqueous 
 solution of caesium bromide, but it nearly all separates on 
 cooling. It is therefore necessary to use a mixture of alcohol 
 and water in preparing this trihalide. A good crop of crystals 
 was obtained by dissolving one-half the theoretical iodine in a 
 solution of one part of caesium bromide dissolved in two parts 
 of water and one part (by volume) of alcohol. 
 
 The following numbers show the composition of the crys- 
 tals: 
 
 P ,, Calculated for 
 
 Found. C8BrIj 
 
 Caesium 28.54 28.48 
 
 Bromine 18.11 17.13 
 
 Iodine 52.01 54.39 
 
 CsBrJ. 
 
 This may be made by dissolving the theoretical amounts of 
 iodine and bromine in a solution of one part of caesium bro- 
 mide in ten parts of water. A considerable excess of bromine 
 does not interfere with its formation. 
 
 The crystals have the following composition : 
 
24 ON A SERIES OF 
 
 w,,r,^ Calculated for 
 
 Found. CflBrjI 
 
 Caesium ..... 31.32 31.67 
 
 Bromine ..... 37.63 38.09 
 
 Iodine ...... 29.57 30.24 
 
 The solubility of this substance in water was approximately 
 determined by estimating the free halogens volumetrically in 
 the mother-liquor from a recrystallization at about 20. The 
 amount found corresponded to 4.45 per cent of CsBr 2 L 
 
 Repeated attempts to make this substance, by using concen- 
 trated solutions of caesium chloride and iodine in mixtures of 
 water and alcohol and cooling to low temperatures, invariably 
 failed. 
 
 CsClBrl. 
 
 This may be made by dissolving about one-fourth of the 
 theoretical bromine and iodine in a solution of one part of 
 caesium chloride in five parts of water. If an excess of 
 caesium chloride is not taken, the product will contain too 
 little chlorine and too much bromine. 
 
 An analysis of the product, properly prepared, gave : 
 
 Vrm-nA Calculated for 
 
 CsClBrl. 
 
 Caesium . . . * . 34.24 35.42 
 
 Chlorine ..... 9.36 9.45 
 
 Bromine ..... 19.96 21.30 
 
 Iodine ...... 32.36 33.83 
 
 If a large excess of bromine is used, an impure product 
 results, as is shown by the following analysis of a sample thus 
 made: 
 
 Found. 
 
 Caesium ........ 36.11 
 
 Chlorine ........ 9.36 
 
 Bromine ........ 27.70 
 
 Iodine ......... 24.83 
 
 On attempting to recrystallize the CsClBrl, a product of a 
 darker red color is formed, sometimes accompanied by the 
 
CAESIUM TRIHALIDES. 25 
 
 separation of some iodine. The following analyses were made 
 of products of recrystallization : 
 
 " A " was from a single recrystallization. " B " was recrys- 
 tallized three times, a little alcohol being added the last time 
 to keep iodine in solution. " C " was recrystallized five times, 
 each time with the addition of a large excess of bromine. 
 
 Calculated for Calculated for 
 A. B. C. CsClBrl. 
 
 Caesium, 32.69 33.22 . . . 35.42 31.67 
 
 Chlorine, 3.32 5.02 2.70 9.45 0. 
 
 Bromine, 31.56 28.30 32.50 21.30 38.09 
 
 Iodine, 30.91 31.78 30.99 33.83 30.24 
 
 These analyses show that the recrystallized body approaches 
 CsBr 2 I, but that a part of the chlorine is very tenaciously held. 
 The excess of bromine used in the case " C " had apparently 
 no effect, probably because caesium and iodine were present in 
 equivalent quantities, whereas, in the case previously given, in 
 which an excess of bromine was used in presence of much 
 caesium chloride in preparing the body, there was evidently a 
 contamination with CsClBr 2 . 
 
 CsClJ. 
 
 This body is dimorphous, the form apparently depending 
 upon the presence or absence of a large excess of caesium 
 chloride in the solution from which it crystallizes. 
 
 The rhombohedral variety is usually obtained by adding 
 iodine in the proportion of one atom to one molecule of 
 caesium chloride dissolved in about ten parts of water, then 
 passing chlorine into the liquid, heated nearly to boiling, until 
 the iodine just dissolves, and finally cooling. If an excess of 
 chlorine is used, the body CsCl 4 I is formed, corresponding to 
 KC1 4 I discovered by Filhol.* This new caesium compound 
 will be described in a subsequent article in connection with 
 several other new bodies of the same class. 
 
 The orthorhombic variety of CsCLJ can be obtained by using 
 
 * Michalis, Anorg Chem., iii, 102 ; Gmelin-Kraut, II, 1, 82. 
 
26 ON A SERIES OF 
 
 three or four times as much caesium chloride as in the other 
 case, the other conditions remaining unchanged. 
 
 The following analyses of two separate products of each of 
 the two varieties were made : 
 
 Rhombohedral. 
 Found. 
 
 Orthorhombic. 
 Found. 
 
 Calculated for 
 CsCl 2 I. 
 
 Caesium, 
 Chlorine, 
 Iodine, 
 
 39.20 
 20.72 
 37.81 
 
 39.92 
 21.08 
 38.21 
 
 3SA3 
 19.78 
 38.97 
 
 40.00 
 20.75 
 38.88 
 
 40.18 
 21.45 
 38.37 
 
 On recrystallizing either form of this substance, from solu- 
 tion in hot water, the rhombohedral variety is usually formed, 
 owing to the lack of an excess of csesium chloride. It is not 
 unusual, however, to obtain at first, as the solution cools, slen- 
 der needles evidently of the orthorhombic variety, which 
 afterward become surrounded by rhombohedral crystals. 
 
 To make this substance, one-half the calculated amount of 
 bromine is added to a solution of one part of csesium bromide 
 in three parts of water, the whole is heated with vigorous 
 shaking until the liquid bromine disappears, and then slowly 
 cooled. Crystals gave on analysis : 
 
 Calculated for 
 Found. CsBr 3 . 
 
 Caesium ...... 35.12 35.66 
 
 Bromine ..... 61.53 64.34 
 
 In preparing this body there usually remains, when the 
 liquid bromine disappears on heating the solution, a heavy red- 
 dish liquid much lighter in color than bromine. It is without 
 doubt a higher polybromide, and it probably corresponds to 
 the easily fusible substance already mentioned as a probable 
 higher polyiodide. An investigation of its composition will 
 soon be made. 
 
 CsClBr z . 
 
 The formation of this body was mentioned at the beginning 
 of this article in connection with the discovery of the new 
 
CESIUM TRIHALIDES. 27 
 
 series of salts. It can be made by adding about one-half the 
 theoretical bromine to a solution of caesium chloride in about 
 five parts of water, dissolving by heat and cooling. 
 
 The analyses of two products are given below. The " pre- 
 cipitate " resulted from adding bromine to a cold csesium 
 chloride solution, and, being finely divided as well as the most 
 unstable compound of the series, it suffered a considerable 
 amount of decomposition, although time was not taken to dry- 
 it thoroughly. 
 
 Found< Calculated for 
 
 
 Precipitate. 
 
 Crystals. 
 
 CsClBr 2 , 
 
 Caesium, 
 
 40.62 
 
 42.14 
 
 40.49 
 
 Chlorine, 
 
 12.64 
 
 13.24 
 
 10.81 
 
 Bromine, 
 
 39.61 
 
 42.93 
 
 48.70 
 
 Water, 
 
 6.45 
 
 1.72 
 
 0. 
 
 CsCl.Br. 
 
 This substance may be made by adding the calculated 
 amount of bromine to a solution of csesium chloride in five 
 parts of water, warming enough to keep CsClBr 2 in solution, 
 and passing chlorine in excess. 
 
 Analysis gave 
 
 Caesium ...... 46.25 46.83 
 
 Chlorine ..... 24.15 25.00 
 
 Bromine ..... 26.05 28.17 
 
 Os C1 8 . 
 
 Efforts to prepare this substance by passing chlorine into 
 saturated aqueous solutions of csesium chloride, cooled by a 
 freezing-mixture, did not succeed. It was noticeable that the 
 compound C1.5H 2 O was not formed under these circumstances. 
 
 Other Trihalides. 
 
 Johnson's KI 8 * is undoubtedly analogous to the csesium 
 compounds. An investigation of this and other potassium 
 
 * J. Chem. Soc., 1877, i, 249. 
 
28 02V A SERIES OF 
 
 trihalides, as well as the corresponding rubidium compounds, 
 is now in progress in this laboratory, and an effort will be made 
 to prepare similar compounds with still other metals. 
 
 The compound KI 2 (CN)* and the body NH 4 I 8 , which John- 
 son describes f as very similar to KI 8 , should be mentioned in 
 this connection. 
 
 There have been described a great number of trihalides of 
 both natural and artificial organic bases. These are mostly 
 triiodides, but there are among them a number of tribromides 
 and also mixed trihalides, especially of the type RC1 2 I. These 
 organic bodies^ are evidently analogous to the caesium com- 
 pounds under consideration, but since they have not been 
 sufficiently studied to throw any light on the structure or crys- 
 talline form of trihalides in general, they will not be mentioned 
 in detail here. 
 
 Theoretical Considerations. 
 
 Thus far in this article, the simplest possible formula? have 
 been used. The probable structure of the compounds will 
 now be discussed. 
 
 The trihalides previously known have been usually con- 
 sidered as weakly combined addition-products. Mendele*eff, 
 for example, says that the instability with which I 2 unites with 
 KI and N(CH 8 ) 4 I is analogous to the instability of many 
 cryohyd rates, e. g., HC1.2H 2 O. It must be noticed that 
 some of the csesium trihalides are very stable, but this fact is 
 evidently due to the strong positive character of the metal, and 
 since others among them are comparatively unstable, it can 
 have no important bearing on their structure. 
 
 Johnson || advances the formula K 2 T 6 for potassium triiodide 
 
 * Langlois, Ann. Chim. Phys., [Ill] Ix, 220. t J. Chem. Soc., 1878, 397. 
 
 J See the following articles: Weltzien, Ann. Ch. Ph., xci, 33; xcix, 1. 
 Miiller, ibid., cviii, 5. Hubner, ibid., ccx, 368. Ladenburg, ibid., ccxvii, 122. 
 Zincke and Lawson, ibid., ccxl, 123. Zincke and Artzberger, ibid., ccxlix, 366. 
 Jb'rgensen, J. pr. Ch., II, vols. ii, iii, xiv, and xv. Dafert, Monatshefte, iv, 496. 
 Dittmar, Berichte, xviii, 1612. Ostermayer, ibid., xviii, 2298. Kamensky, 
 ibid., xi, 1600. Tilden, J. Chem. Soc., xviii, 99. Hoogewerff and Dorp, Rec. 
 Trav. Chim., iii, 361. 
 
 Grundlagen der Chemie, p. 563, foot-note 63. 
 
 || J. Chem. Soc., 1878, 183. 
 
CAESIUM TRIHALIDES. 29 
 
 with no better reason than the existence of a higher iodide of 
 mercury, HgI 6 .* This formula may be dismissed at once, for 
 there is no more ground for it than for writing K 2 I 2 because 
 HgI 2 exists, and moreover, if the caesium salts were Cs 2 X 6 , we 
 should expect to find in the series such compounds as Cs 2 Cl 6 I, 
 Cs 2 Cl 8 l8, etc., none of which were discovered. 
 
 Since the members of the caesium series are crystallographi- 
 cally isomorphous, they must all have the same structure, and, 
 as has just been shown, no multiple of the formula CsX 8 is 
 probable. 
 
 A possible explanation of the caesium trihalides may be 
 made by supposing the metal to act trivalently, and the fol- 
 lowing arguments seem to favor this view : 
 
 1st. Caesium has the highest atomic weight of the alkali- 
 metals, and it is a noticeable fact that, among the elements 
 in general, those with higher atomic-weights in a group have 
 the greater tendency to act with variable quantivalence. 2d. 
 Caesium is univalent in its ordinary compounds, and, following 
 the general rule, the next higher quantivalence should be 
 three. 3d. Caesium is in the same group as gold in Men- 
 dele'eff's periodic system of the elements, and it is well known 
 that this element acts univalently and trivalently. 
 
 On the other hand, the following arguments are in favor of 
 considering the bodies double salts : 1st. The compounds IBr, 
 IC1, and BrCl are definite bodies ; and all the trihalides may be 
 considered as molecular compounds of these, and also the 
 molecules L and Br 2 , with normal halides. 2d. The fact that 
 CsBr 2 I is more stable than CsBrI 2 , and that CsCl 2 Br is more so 
 than CsClBr^ showing that the stability of these bodies does 
 not entirely depend upon the volatility of the halogens con- 
 tained in them, indicates that the halogen atoms have much 
 influence on each other, and that at least two of them are 
 probably bound together. A consideration of the fact that 
 CsCl 2 I is a very stable body, while CsClLz probably cannot be 
 prepared, leads to the same conclusion. 3d. It has been pointed 
 out by Godeffroy f that the simple salts of caesium are as a rule 
 * Jorgensen, J. pr. Ch., II, ii, 357. t Berichte, ix, 1365. 
 
30 ON A SERIES OF 
 
 more soluble than the corresponding rubidium and potassium 
 salts, while with the double salts the reverse is true. Work 
 now in progress in this laboratory shows that the rubidium and 
 potassium trihalides, as far as they have been investigated, 
 increase in solubility towards potassium ; hence, if the rule 
 holds true, they must be double salts. 4th. The new salt 
 CsI.AgI,* an undoubted double salt, shows a close relation to 
 the trihalides in its system of crystallization, in the ratio of 
 the two axes which alone were determined, and in its cleav- 
 age.f The salt KI.AgI J has been prepared, but it has not yet 
 been procured in crystals fit for measurement. Work will be 
 continued on this class of compounds. 
 
 The evidence which has just been given points strongly 
 towards considering the trihalides as double salts. The con- 
 siderations based on the axial relations of the crystals, which 
 have been given in the crystallographic part of this article, 
 also indicate that these bodies cannot be viewed as compounds 
 of trivalent caesium, and, moreover, that a single halogen 
 atom of highest atomic weight plays an important part in their 
 structure. One view of the possible position of this peculiar 
 atom may be that it is a trivalent atom united to the others in 
 
 the manner indicated by the general formulae Cs I<v and 
 
 * This was made by dissolving Agl in a concentrated, hot solution of Csl 
 and cooling. It forms tufts of hair-like crystals, very seldom large enough to 
 measure. 
 
 Analysis gave C ^SS^ U 
 
 Silver 19.03 21.82 
 
 Iodine 50.91 51.32 
 
 Caesium 26.86 
 
 t CsI.AgI crystallizes in the orthorhombic system. It was obtained in long 
 needles, not over mm. in diameter and too small to show distinct end faces. 
 A pinacoid and two other forms occur in the prismatic zone which correspond 
 to c(001), #(012), and c?(011) of the trihalide series. Of these c and g are the 
 best developed, while d is very small. Five independent measurements of 
 c A g gave values varying between 27 29' and 27 44', the average being 27 36', 
 which gives the ratio b : c = 1 : 1.0456. The crystals show an imperfect cleav- 
 age parallel to a (100) and probably a second parallel to c, but they were too 
 small to allow of this being determined with certainty. Under the polarizing 
 microscope they show parallel extinction. 
 
 \ Boullay, Ann. Chim. Phys., II, xxxiv, 377. 
 
CAESIUM TRIHALIDES. 
 
 31 
 
 Cs Br<-^. This view may be objected to because the 
 strongest halogen atom is not directly united to the caesium. 
 
 V 
 
 The structure Cs X< n is scarcely admissible in the com- 
 
 X 
 
 pound CsI.AgI on account of the probable invariable univa- 
 lence of silver. 
 
 Another view of the position of the peculiar iodine or 
 bromine atom may be based upon the idea, so ably advocated 
 by Remsen,* that a bivalent group of two halogen atoms acts 
 the same part in double halides that oxygen takes in ordinary 
 oxygen salts. Supposing that a halogen atom of highest atomic 
 weight is linked to the caesium atom by means of the other 
 two halogen atoms, acting as a bivalent group, the trihalides 
 have the following formulae : 
 
 Cs (II) I 
 Cs (BrI) I 
 Cs (BrBr) I 
 Cs (ClBr) I 
 Cs (C1C1) I 
 [Cs-(ClI) I] 
 
 Cs (BrBr) Br 
 Cs (ClBr) Br 
 Cs (C1C1) Br 
 
 [Cs (C1C1) Cl] 
 
 These formulae allow the supposition that the most negative 
 halogen atom is in direct union with the caesium. They are 
 consistent with the view that there are two symmetrical series 
 of compounds, one with an iodine atom, the other with a 
 bromine atom in a special position. On these grounds they 
 seem quite plausible. 
 
 Assuming that the structure of the trihalides is represented 
 by the above formulae, a comparison of their relative stability 
 leads to the view that the linking group of two halogen atoms 
 causes greater stability when composed of like atoms than 
 when these atoms differ. The groups (II) (BrBr) and (C1C1) 
 occur in the comparatively stable compounds, while (ClBr) is 
 in the most unstable body of the " bromine series," (BrI) is in 
 
 * On the Nature and Structure of Double Halides, Amer. Chem. Jour., 
 xi, 291. 
 
32 ON A SERIES OF CAESIUM TRIHALIDES. 
 
 the least stable one of the " iodine series," and the compound 
 which should contain (C1I), with the least closely related halo- 
 gens, could not be prepared. There is a possible exception in 
 the body Cs (ClBr) I, for it has not been noticed that the 
 stability of this varies to any marked extent from Cs (C1C1) 
 I. The same view of the effect of the identity of the link- 
 ing halogen atoms will probably apply to all double halides, 
 for it is certain that very few of these containing different 
 halogens are known, although this may be partly due to the 
 fact that mixed double halides have not been sufficiently 
 studied. An investigation of the double halides of caesium 
 and mercury, now in progress, indicates that the generalization 
 will apply in this case. Whatever may be the true structure 
 of double halides, this influence of the identity of the halogens 
 in the combining simple salts probably depends upon the same 
 broad chemical law which causes, for example, two oxides or 
 two sulphides to combine more readily than an oxide and a 
 sulphide of the same elements, and which causes sulphates to 
 combine with other sulphates more readily than they combine 
 with nitrates and other dissimilar salts. 
 
 The present communication may be considered as prelimi- 
 nary to a wider study of polyhalides and double halogen salts. 
 It is hoped that other series, studied chemically and crystallo- 
 graphically, may give valuable results. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 December, 1891. 
 
ON THE RUBIDIUM AND POTASSIUM 
 TRIHALIDES.* 
 
 BY H. L. WELLS AND H. L. WHEELER. 
 
 WITH THEIR CRYSTALLOGRAPHY. 
 
 BY S. L. PENFIELD. 
 
 THE discovery of a series of csesium trihalides f has led the 
 writers to investigate the analogous rubidium and potassium 
 compounds. The following table gives a list of the bodies 
 which we have been able to prepare, together with the csesium 
 series for comparison. The compound KI.I 2 had been pre- 
 viously prepared by Johnson.J 
 
 CsI.I 2 RbI.I 2 KLI 2 
 
 CsBr.I 2 
 
 CsBr.BrI RbBr.BrI KBr.BrI 
 
 CsCLBrI RbCLBrl 
 
 csci.cn Rbci.cn KCI.CII 
 
 CsBr.Br 2 RbBr.Br 2 
 
 CsCl.Br 2 RbCl.Br 2 
 
 CsCl.ClBr RbCl.ClBr 
 
 It is to be noticed that there is but one member lacking in 
 the rubidium series to make it as complete as that of caesium. 
 We have repeatedly tried to prepare this compound, RbBr.I 2 , 
 using alcoholic solutions of varying strength and great con- 
 centration at low temperatures, but with no success. The 
 failure to make this body doubtless depends upon the com- 
 parative instability of the rubidium series. We have even 
 attempted to prepare RbCl.I 2 and RbCl.Cl 2 , corresponding 
 
 * Amer. Jour. Sci., xliv, July, 1892. 
 t Ibid., Ill, xliii, 17. 
 J J. Chem. Soc., 1877, 241. 
 3 
 
34 ON THE RUBIDIUM AND 
 
 to which no caesium compounds could be made, but, as was 
 anticipated, these efforts were entirely without success. 
 
 In the potassium series, only those bodies could be prepared 
 which correspond to the more stable caesium and rubidium 
 compounds. They show a great decrease in stability in 
 comparison with the rubidium compounds. A product was 
 obtained at a very low temperature, which was probably KBr. 
 Br 2 , but we did not make a satisfactory analysis of it. 
 
 We have attempted to prepare a number of sodium and 
 lithium trihalides. There is no doubt that some of them 
 exist, but they are so extremely soluble and unstable that we 
 have abandoned work in this direction. 
 
 Method of Preparation. The rubidium and potassium 
 compounds are made, like the caesium series, by dissolving a 
 normal halide with the proper halogen or halogens in water 
 with the aid of heat and cooling to crystallization. The 
 members of the rubidium series, being very soluble, require 
 very concentrated solutions for their preparation. The potas- 
 sium compounds, being still more soluble, require the greatest 
 possible degree of concentration, and are usually best obtained 
 by exposing the solutions for a considerable time to a winter 
 temperature, evaporation in the desiccator being sometimes 
 also necessary. 
 
 Color. The colors of the rubidium and potassium com- 
 pounds are very similar to those of the corresponding members 
 of the caesium series, but, since they usually form larger 
 crystals, their apparent color is generally somewhat darker. 
 They vary in color from brilliant black in RbI.I 2 and KI.I 2 , 
 through various shades of yellowish-red and orange to bright 
 yellow in the compound RbCl.ClBr. In all the compounds 
 that have been prepared, the color becomes lighter as the sum 
 of the atomic weights of the three halogen atoms decreases. 
 
 Stability. It has been found by experiment that the 
 potassium trihalides are much less stable on exposure to the 
 air than the corresponding rubidium compounds, while these 
 in turn are less stable than the members of the caesium series. 
 The same relative stability of the three series is shown by the 
 
POTASSIUM TRIHALIDES. 35 
 
 temperatures at which they are completely decomposed by 
 rapid heating as given below : 
 
 Approximate Temperature of Whitening. 
 
 CsI.I 2 , 330 
 CsBr.Brl, 320 
 CsCLClI, 290 
 CsCLBrl, 290 
 
 KbI.I 2 , 270 
 EbBr.BrI, 265 
 BbCLClI, 265 
 BbCLBrl, 200 
 
 KI.I 2 , 225 
 KBr.BrI, 180 
 KC1.C1I, 215 
 
 CsBr.Br 2 , 160 
 
 KbBr.Br 2 , 140 
 
 
 CsCl.ClBr, 150 
 
 BbCLClBr, 110 
 
 
 CsCLBr., 150 
 
 BbCUBr,, 80 
 
 
 Fusibility. The melting-points of the analogous com- 
 pounds become lower from caesium to potassium. In the 
 open capillary tube RbI.I 2 melts at 194 and RbCl.ClI at 
 208, while all the other rubidium compounds whiten without 
 melting. The potassium compounds give practically the 
 same melting-points in open as in sealed tubes. The follow- 
 ing table gives the approximate melting-points in sealed 
 tubes : 
 
 CsLI,, 201-208 KbIJ 2 , 190 KI.I 2 , 38* 
 
 CsBr.Brl, 243-248 KbBr.BrI, 225 KBr.BrI, 60 
 
 CsCl.ClI, 225 -230 BbCl.CH, 180 -200 KC1.C1I, 60 
 
 CsCLBrl, 225-235 BbCLBrl, 205 
 
 CsBr.Br 2 , 180 EbBr.Br 2 , whitens 
 
 CsCl.ClBr, 205 BbCLClBr, whitens 
 
 CsCl.Br 2 , 191 KbCl.Br 2 , 76? 
 
 Behavior with Solvents. The extreme solubility of the 
 rubidium and potassium trihalides in water has already been 
 referred to, and it has been pointed out that the members of 
 the potassium series are the most soluble. The rubidium 
 compounds which contain iodine can be recrystallized from 
 water without difficulty. These four bodies containing rubid- 
 ium and iodine are sufficiently stable to be soluble in alcohol, 
 while the remaining rubidium compounds, as well as all the 
 
 * Johnson gives 45 for the melting-point of this compound (1. c.). 
 
36 ON THE RUBIDIUM AND 
 
 potassium compounds, are more or less readily decomposed 
 by alcohol with the separation of normal halides. Ether 
 decomposes all the rubidium and potassium compounds, leav- 
 ing normal halides undissolved. 
 
 Crystallography. 
 
 The rubidium trihalides crystallize in the orthorhombic 
 system and are isomorphous with the corresponding csesium 
 compounds, showing a close similarity both in crystalline 
 habit and in axial ratios. 
 
 The forms which have been observed are : 
 
 a, 100, i-1 d, Oil, 1-r 
 
 b, 010, i-i f, 021, 2-r 
 
 c, 001, e, 102, \-l 
 m, 110, 1 p, 111, 1. 
 
 With the exception of the pyramid p, which was observed 
 as a small face only on RbI.I 2 , these are the same as were 
 observed on the csesium trihalides, while the brachydome g, 
 012, J-i, which was found only on CsI.I 2 , was not observed on 
 any of the rubidium compounds. 
 
 Of the three potassium trihalides which were examined, 
 only one, KBr.BrI, was orthorhombic like the caesium and 
 rubidium compounds. The others, KI.I 2 and KC1.C1I, are 
 monoclinic, but they can be referred to axes which are similar 
 to those of the orthorhombic series. 
 
 The cleavage of the rubidium trihalides is perfect parallel to 
 c, less perfect parallel to a ; neither is easily produced. The 
 crystals are very brittle and usually break with a conchoidal 
 fracture. The potassium trihalides are exceedingly brittle, 
 and no cleavage was observed. The optical properties were 
 not studied, owing to the difficulty of preparing orientated 
 sections. 
 
 In the following table the axial ratios of all the alkali- 
 metal trihalides are given, arranged as in the csesium paper. 
 In the table of angles those which were chosen as fundamen- 
 tal are marked by an asterisk. 
 
POTASSIUM TRIHALIDES. 
 
 37 
 
 ii. 
 
 
 a : b : c 
 
 a : b : c 
 
 
 Series with iodine. 
 
 
 ( CsI.I 2 
 J EbU, 
 
 0.6824 : 1 : 1.1051 
 0.6858 : 1 : 1.1234 
 
 1 
 1 
 
 1.4655 
 1.4582 
 
 1.6196 
 1.6381 
 
 (KLI, 
 
 ( 0.7065 : 1 : ... 
 
 1 
 
 1.4154 
 
 
 
 | Monoclinic, a = 86 47' 
 
 a = 86 
 
 47$' 
 
 CsBr.I 2 
 
 0.6916 : 1 : 1.1419 
 
 1 
 
 1.4460 
 
 1.6511 
 
 f CsBr.BrI 
 
 0.7203 : 1 : 1.1667 
 
 1 
 
 1.3882 
 
 1.6196 
 
 } EbBr.Brl 
 
 0.7130 : 1 : 1.1640 
 
 1 
 
 1.4025 
 
 1.6325 
 
 ( KBr.BrI 
 
 0.7158 : 1 : 1.1691 
 
 1 
 
 1.3970 
 
 1.6333 
 
 ( CsCLBrl 
 
 0.7230 : 1 : 1.1760 
 
 1 
 
 1.3831 
 
 1.6268 
 
 t EbCLBrl 
 
 0.7271 : 1 : 1.1745 
 
 1 
 
 1.3753 
 
 1.6153 
 
 ( CsCl.ClI 
 J EbCl.ClI 
 
 0.7373 : 1 : 1.1920 
 0.7341 : 1 : 1.1963 
 
 1 
 
 1 
 
 1.3563 
 1.3622 
 
 1.6167 
 1.6296 
 
 (Kci.cn 
 
 ( 0.7335 : 1 : 1.2204 
 
 1 
 
 1.3633 
 
 1.6638 
 
 1 Monoclinic, a = 83 20' a = 83 20' 
 
 
 Series without iodine. 
 
 
 ( CsBr.Br 2 
 
 0.6873 : 1 1.0581 
 
 1:1.4550 1.5395 
 
 ( EbBr.Br 2 
 
 0.6952 : 1 1.1139 
 
 1:1.4384 1.6023 
 
 (CsCLBr. 
 
 0.699 : 1 ... 
 
 1 : 1.430 
 
 (EbCl.Br 2 
 
 0.70 : 1 1.1269 
 
 1 : 1.43 1.61 
 
 JCsCLClBr 
 
 0.7186:1 1.1237 
 
 1:1.3917 1.5638 
 
 i EbCl.ClBr 
 
 0.7146:1 1.1430 
 
 1 : 1.3994 1.5995 
 
 
 m A m, 110 A fiO d A d, Oil 
 
 A Oil e A c, 102 A 102 
 
 EbI.I 2 
 
 *68 53' *96 
 
 39' 78 38' 
 
 KI.I 2 
 
 *70 34' 
 
 
 EbBr.Brl 
 
 70 58' *98 
 
 40' *78 27' 
 
 KBr.BrI 
 
 71 12' *98 
 
 55' 78 28' 
 
 EbCLBrl 
 
 72 2' *99 
 
 10' *77 51' 
 
 EbCl.ClI 
 
 72 34' *100 
 
 13' *78 21' 
 
 KCi.cn 
 
 72 54' 
 
 *79 8' 
 
 EbBr.Br 2 
 
 *69 37' *96 
 
 10' 77 24' 
 
 EbCLBr 2 
 
 *70 approx. *96 
 
 58' 76 appro 
 
 EbCl.ClBr 
 
 71 6 f *97 
 
 38' *77 18' 
 
 A comparison of the axial ratios of the trihalides shows 
 that the replacement of csesium by rubidium, and in one case 
 
38 ON THE RUBIDIUM AND 
 
 by potassium, has little or no effect on the form, while in two 
 of the compounds potassium causes a change in symmetry 
 without much change in the axes. It is evident that the 
 rubidium salts, like those of caesium, may be arranged in two 
 symmetrical series, one with and the other without iodine, in 
 which the ratio of two axes remains nearly constant through- 
 out, while the third varies, and the conclusions which were 
 arrived at in our previous paper concerning the constitu- 
 tion of the caesium trihalides are confirmed by the rubidium 
 compounds. 
 
 The rubidium trihalides have a strong tendency to crystal- 
 lize, and the solubility is such that, from solutions of not over 
 50 c. c. in volume, large and magnificent crystals several cen- 
 timetres in length, can readily be obtained. The size of the 
 crystals seems often dependent only upon the volume of the 
 solution and the size of the vessel containing it. Many of 
 the large crystals are complex, being built up of smaller ones 
 in parallel position. Some of the crystallizations were as 
 beautiful as any that we have ever seen. 
 
 The rubidium trihalides containing iodine were measured 
 at ordinary temperatures ; those without iodine and the potas- 
 sium trihalides at about C. It was found that the stability 
 of the compounds increased very rapidly with a diminution 
 in temperature and, by working in the cold, no difficulty was 
 experienced in making accurate measurements of the more 
 unstable salts. It is not considered necessary to give with 
 each trihalide a table of measured and calculated angles, but 
 in all cases where a series of accurate measurements were 
 obtained, they agreed closely with the calculated. 
 
 EbI.I 2 . The forms 6, c, m, d, f, e, and p were observed. 
 Of these/ and p were always small and frequently wanting. 
 The habit is shown in Fig. 1. 
 
 RbBr.BrL The forms of a, c, m, d, and e were observed. 
 The habit is shown in Fig. 2. 
 
 KbOl.Brl. The forms a, d, and e were observed. The 
 pinacoid a is usually wanting, and the simple habit shown in 
 Fig. 3 prevails. 
 
1. 
 
 POTASSIUM TRIHALIDES. 
 2. 
 
 39 
 
 ThQ forms a, d, and e were observed. The 
 habit is shown in Fig. 4. 
 
 RbBr.Br*. The forms a, b, m, d, and / were observed. 
 The habit is shown in Fig. 5. 
 
 RbCl.Br*. The forms 5, c, m, c?, and e were observed. 
 The habit is shown in Fig. 6. The tendency of this salt is 
 to crystallize in small scales ; it is also the most unstable of 
 the rubidium series, so that we considered it fortunate that 
 
 we were able to make out the axial ratio. The only faces 
 which yielded good reflections were b and c?, which established 
 accurately the relation between the b and c axes, while only 
 approximate measurements were obtained from the other faces. 
 
 RbCl.ClBr. The forms a, 5, m, d, and e were observed. 
 The habit is like Fig. 2, except that c is wanting. 
 
 KHz. This occurs in very simple monoclinic crystals. 
 If the solution is cooled slowly it forms in stout prisms, but 
 by rapid cooling a network of fine needles is obtained. In 
 order to make this salt and the monoclinic KC1.C1I conform 
 to the position which has been adopted for the orthorhombic 
 trihalides, it is necessary to deviate from the ordinary custom 
 
40 
 
 ON THE RUBIDIUM AND 
 
 and make the clino-axis slope from right to left instead of 
 from back to front. The faces are taken as 6, 010, i-l ; <?, 001, 
 and ?tt, 110, /. The crystals are not sufficiently modified 
 to determine more than two axes, but taking as fundamental 
 measurements, b A 7/1, 010 A 110 = 54 43' and c A c (re-entrant 
 angle of twin crystal) = 6 25', the following axial ratio is 
 obtained, a : b = .7065 : 1 ; a = 010 A 001 = 86 47 J'. The 
 angle m A c, 110 A 001, was measured 91 55' and 91 50', cal- 
 culated 91 51'. Fig. 7 represents a twin crystal in the above 
 position. Fig. 8 represents a simple crystal in the ordinary 
 monoclinic position, with d as the clino axis. The axial ratio 
 for this position is, d:t = 1.4154 : 1 ; = 86 47 J'. 
 
 m 
 
 "s/ 
 
 KBr.BrL The forms , 5, w, c?, /, and e were observed. 
 The habit is shown in Fig. 9. This salt differs from all of 
 the other alkali-metal trihalides in having the brachy prism 
 w, 120, i-2, instead of the unit prism m. The fundamental 
 measurements were a A n 100 A 120 = 55 4' and d A d, Oil 
 A Oil, = 98 55'. 
 
 KCL OIL This crystallizes in long needles belonging to 
 
 the monoclinic system, Fig. 10. 
 Taking b as the clino axis, the 
 forms are a, 100, i-i; 6, 010, i-l; 
 e, 001, ; x, 032, f-i, and , 102, \-i. 
 The measurements taken as funda- 
 mental are c A 5, 001 A 010 = 96 40', 
 e A e, 102 A 102 = 79 8', and c A z, 001 A 032 = 66 35', from 
 
 to. 
 
POTASSIUM TRIHALIDES. 41 
 
 which the following axial ratio was calculated, a : b : c .7335 
 : 1 : 1.2204, a = 83 20'. If taken in the ordinary monoclinic 
 position with e as the prism 110 and, a: as the orthodome 
 101, the axial ratio from the above measurements becomes 
 d:~b:c = .8319 : 1 : .4544; = 83 20'. 
 
 Method of Analysis. 
 
 The methods used for the analyses of the potassium and 
 rubidium trihalides were exactly the same as those men- 
 tioned in the article on caesium trihalides. 
 
 The crystals were prepared for analysis by pressing between 
 papers and at the same time crushing them somewhat. In 
 some cases, where the bodies were very easily decomposed, 
 this was done in cold weather out of doors, but, even with this 
 precaution, it was not possible to dry them very thoroughly or 
 to avoid a considerable amount of decomposition. 
 
 RU.I*. 
 
 This body can be prepared by dissolving 55 g. of rubidium 
 iodide in enough water to make a solution of 50 c. c., adding 
 60 g. of iodine, warming until solution takes place, and cooling 
 to ordinary temperature. A mass of large crystals in parallel 
 position, forming steps, is usually formed. 
 
 Analysis gave <%$$* 
 
 Kubidium .... 18.32 18.32 18.33 
 
 Iodine 81.07 . . . 81.67 
 
 A specific gravity determination, made in the mother-liquor 
 at 22, gave the number 4.03. This cannot be considered very 
 exact, on account of the difficulty of obtaining the mother- 
 liquor in such a condition that it neither dissolves nor 
 deposits the substance. A sample of mother-liquor, of specific 
 gravity 2.19, was found to contain 1.61 g. of RbI.I 2 in 1 c. c. 
 The compound therefore dissolves in about one-third its 
 weight of water at 22. It is interesting to notice here that, 
 under nearly the same conditions, the corresponding caesium 
 compound, CsI.I 2 , requires more than one hundred parts of 
 
42 ON THE RUBIDIUM AND 
 
 water to dissolve it. It is expected that this great difference 
 in solubility will form the basis of a useful method for 
 separating the two metals. 
 
 KbBr.BrL 
 
 This compound can be readily made by dissolving, with the 
 aid of heat, 30 g. of iodine and 20 g. of bromine in a saturated 
 aqueous solution of 40 g. of rubidium bromide and cooling. 
 The facility with which this body crystallizes is remarkable. 
 The large crystals have a color and lustre much like the 
 mineral pyrargyrite, " ruby-silver." 
 
 Analysis 
 
 Kubidium ...... 22.79 22.95 
 
 Bromine ...... 45.19 42.95 
 
 Iodine ....... 31.11 34.10 
 
 An approximate specific gravity determination, made with 
 the mother-liquor, gave the number 3.84. An analysis of the 
 mother-liquor showed that it contained about 44 per cent of 
 RbBr.Brl. The mother-liquor of the corresponding caesium 
 compound contained only 4.45 per cent of CsBr.Brl. 
 
 MCl.BrL 
 
 This body can be made by adding 27 g. of bromine and 
 42 g. of iodine to a saturated aqueous solution of 40 g. of 
 rubidium chloride, warming until all is in solution, and cool- 
 ing. It forms magnificent crystals which can be readily 
 recrystallized from water. Unlike the corresponding caesium 
 compound, it does not change its composition by recrystalliza- 
 tion ; hence it is probable that it is a true chemical compound, 
 and not a mixture of the isomorphous bodies RbBr.Brl 
 and RbCl.CH. 
 
 Analysis gave, Calculated for 
 
 Original crystals. 6th recrystallization. RbCl.Brl. 
 
 Rubidium .... 26.67 27.34 26.06 
 
 Chlorine .... 10.65 . . . 10.82 
 
 Bromine .... 24.89 . . . 24.39 
 
 Iodine 38.13 38.72 
 
POTASSIUM TRIHALIDES. 43 
 
 RbCLCIL 
 
 A convenient method for preparing this compound is to 
 pass chlorine into a warm, concentrated solution of rubidium 
 chloride, containing the calculated amount of iodine, until the 
 iodine is just dissolved. If too much chlorine is used, the 
 compound RbCl 4 I is formed, which we shall describe in a 
 future article. It is best to stop adding chlorine while the 
 solution is still colored red by iodine. On cooling the liquid, 
 the compound separates, usually in large flat groups of parallel 
 crystals. 
 
 Analysis gave 
 
 Rubidium ...... 29.85 30.15 
 
 Chlorine ...... 24.68 25.04 
 
 Iodine . 44.68 44.79 
 
 This can be prepared by adding 49 g. of bromine to 45 c. c. 
 of an aqueous solution containing 50 g. of rubidium bromide, 
 warming gently until bromine dissolves, then cooling. It 
 usually forms a mass of large, brilliant, red crystals in parallel 
 position. 
 
 Analysis gave 
 
 Rubidium ...... 25.86 26.26 
 
 Bromine ...... 73.09 73.73 
 
 This body is prepared by adding bromine to a warm, satu- 
 rated solution of rubidium chloride until some bromine 
 remains undissolved, and cooling to a low temperature. The 
 compound crystallizes well, but it is the most unstable of the 
 seven rubidium trihalides that have been prepared, and, 
 although it was not fully dried, the sample used for analysis 
 suffered a considerable amount of decomposition. 
 
44 ON THE RUBIDIUM AND 
 
 Analysis gave 
 
 Kubidiuin .... 32.57 . . . 30.42 
 
 Chlorine .... 14.46 14.44 12.63 
 
 Bromine .... 49.04 49.40 56.93 
 
 In one attempt to prepare this trihalide too much water was 
 used, and it was necessary to evaporate off the bromine and 
 concentrate the solution. This operation was repeated several 
 tunes, after fresh additions of bromine, before the proper con- 
 ditions were arrived at, and the product finally obtained was 
 contaminated with RbBr.Br 2 , as is shown by the following 
 analyses : 
 
 Wnnnd Calculated for Calculated for 
 
 RbCl.Br 2 . RbBr.Brj. 
 
 Kubidium . . . 28.78 . . . 30.42 26.26 
 
 Chlorine . . . 7.66 6.94 12.63 0. 
 
 Bromine . . . 60.92 61.37 56.93 73.73 
 
 We have found by experiment that rubidium chloride is 
 partly changed to bromide by evaporating an aqueous solution 
 of it with bromine.* This explains the formation of the 
 RbBr.Br 2 . 
 
 EbCl.ClBr. 
 
 This compound can be prepared by adding 33 g. of bromine 
 to a saturated solution of 50 g. of rubidium chloride, passing 
 chlorine to saturation into the slightly warmed solution, and 
 cooling to a low temperature. The substance is usually 
 deposited in the form of very large, light yellow prisms. 
 
 A , . ^ Calculated for 
 
 Analysis gave RbCl.ClBr. 
 
 Rubidium .... 35.42 35.41 36.15 
 
 Chlorine .... 29.27 28.96 30.02 
 
 Bromine .... 31.56 31.39 33.82 
 
 KLI*. 
 
 This body can be made in a few hours by dissolving the 
 theoretical amount of iodine in a hot, saturated, aqueous solu- 
 
 * This is in accordance with the results of Potilzin referred to by Men- 
 dele'eff in his "Grundlagen der Chemie" (German ed., 1891), p. 638. 
 
POTASSIUM TRIHALIDES. 45 
 
 tion of potassium iodide and exposing the resulting solution to 
 a winter temperature. It can also be made, as Johnson states,* 
 by evaporating the solution in a desiccator for a long time. 
 Johnson states that he always obtained a crop of potassium 
 iodide before the triiodide separated. We have never obtained 
 such a product, undoubtedly because we have invariably used 
 a sufficient amount of iodine. 
 
 It was not considered necessary to make a new analysis of 
 this body. 
 
 KBr.BrL 
 
 This compound can be prepared by making a very concen- 
 trated, warm solution of the calculated amounts of potassium 
 bromide, bromine, and iodine, and exposing it for some time to 
 a low temperature. The product used for analysis was well 
 crystallized, but it suffered rapid decomposition on exposure 
 to the air. 
 
 Analysis gave ^Srl^ 
 
 Potassium .... 12.21 . . . 11.99 
 
 Bromine .... 51.25 51.61 49.06 
 
 Iodine 30.42 29.11 38.94 
 
 KCl.CIL 
 
 To prepare this substance, chlorine is passed into a warm 
 mixture of calculated quantities of potassium chloride and 
 iodine, in the presence of an amount of water insufficient to 
 dissolve the potassium chloride even when hot. The stream 
 of chlorine is stopped as soon as the iodine has been converted 
 into the monochloride, for otherwise Filhol's well-known com- 
 pound KC1.C1 3 I will be formed. Everything is then dis- 
 solved by warming and cautiously adding water if necessary, 
 and the solution is exposed to a low winter-temperature. The 
 crystals are very unstable, but apparently not quite as much so 
 as KBr.BrL 
 
 Analysis gave *$88& a 
 
 Potassium .... 15.29 15.35 16.49 
 
 Chlorine .... 27.53 27.50 29.94 
 
 Iodine 50.37 50.12 53.56 
 
 * Loc. cit. 
 
46 ON THE RUBIDIUM AND 
 
 Other Double Halides. 
 
 The double salt CsI.AgI was described in connection with 
 the caesium trihalides as being isomorphous with them as far 
 as the crystals could be measured. Much work has since been 
 done, without avail, in the hope of obtaining better crystals of 
 this compound. Unsuccessful efforts have been made to 
 obtain measurable crystals of all the corresponding silver 
 double halides (except the fluorides) with caesium, rubidium, 
 and potassium. Two or three of these compounds had already 
 been described, and it is probable that we could have proven 
 the existence of all the rest of them, but the poorly crystal- 
 lized products obtained had no interest in this connection and 
 were not analyzed. Repeated efforts also failed to produce 
 from potassium iodide and cuprous iodide a double salt that 
 could be measured. 
 
 Theoretical. 
 
 Arguments were given in the article on the caesium series 
 which have led us to regard the trihalides as belonging to the 
 class of bodies called double halides. We have indicated this 
 view in the present article by using the usual formulae for 
 such compounds. 
 
 The well-known idea of a linking group of two halogen 
 atoms as an explanation of the structure of double halides was 
 advocated for the caesium trihalides, and, since the rubidium 
 and potassium compounds are entirely analogous, it is unneces- 
 sary to give their structural formulae here. We believe, how- 
 ever, that the trihalides throw some light upon the constitution 
 of the diatomic linking group. Remsen says,* " I cannot see 
 that at present we have any evidence which justifies us in the 
 use of the expression C1 = C1 rather than Cl Cl ." 
 If, as we believe, the structure of rubidium triiodide is ex- 
 pressed by the formula Rb (II) I, the structure of the link- 
 ing group probably cannot be I I ; for in that case a 
 single bivalent iodine atom could do the linking as well as a 
 group of two, and we should expect the existence of diiodides, 
 
 * Araer. Chem. Jour., xi, 312. 
 
POTASSIUM TRIHALWES. 47 
 
 no evidence of which, or of any other dihalide, has been found 
 in the course of an elaborate investigation of the alkali-metal 
 polyhalides. Moreover, with the assumption of bivalent halo- 
 gen atoms, there would be no difficulty in supposing four 
 halogens to be linked together, and the existence of tetra- 
 halides would be anticipated. Our investigations, however, 
 have shown the existence of only tri- and pentahalides.* The 
 double linking seems therefore the more probable of the two 
 forms mentioned by Remsen, but it may be added that any 
 union weaker or stronger than the others in the molecule, and 
 different from them, would also explain the non-existence of 
 dihalides and tetrahalides. 
 
 Assuming that there is a linking group of two halogen 
 atoms in the trihalides, the view advanced from a considera- 
 tion of the caesium compounds, that the most stable bodies 
 have identical atoms in this group, is confirmed by the study of 
 the rubidium and potassium analogues. For, on this assump- 
 tion, all the potassium compounds which could be made con- 
 tain a group of identical atoms, while in the missing rubidium 
 compound they are dissimilar. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 March, 1892. 
 
 * The pentahalides will be described in a future article. 
 
ON THE ALKALI-METAL PENTAHALIDES.* 
 
 BY H. L. WELLS AND H. L. WHEELER. 
 
 WITH THEIR, CRYSTALLOGRAPHY. 
 
 BY S. L. PENEIELD. 
 
 IN the course of our investigations on the alkaline trihalides,f 
 the compounds CsCl.ClJ, RbCl.Cl 3 I, and KC1.C1 3 I were 
 encountered. The potassium compound had been described 
 many years ago by Filhol.J This investigator prepared also 
 the body NH 4 C1.C1 8 I and obtained a similar magnesium 
 compound, probably MgCl 2 .2Cl 8 I.5H 2 O. He failed in his 
 attempts to make analogous compounds with sodium and a 
 considerable number of the other common metals. 
 
 It was evident from the peculiar behavior of caesium tribro- 
 mide and triiodide, mention of which was made in one of our 
 previous articles, that a still higher bromide and iodide 
 existed. These have now been identified as pentahalides. 
 
 In addition to these bodies, we have prepared the sodium 
 and lithium analogues of Filhol's salt. They differ from all 
 the other polyhalides that we have studied, in containing water 
 of crystallization. 
 
 A large number of other alkaline pentahalides are theoreti- 
 cally possible, but, although we have made numerous experi- 
 ments with the view of making the most promising of these, 
 we have been unable to prepare them. It may be stated that 
 special efforts were made to obtain potassium and rubidium 
 pentaiodides. 
 
 * Amer. Jour. Sci., xliv, July, 1892. 
 
 t Ibid, III, xliii, 17 and 475. 
 
 J J. Pharm., xxv (1839), 431. 
 
 Amer. Jour. Sci., Ill, xliii, 24 and 27. 
 
ALKALI-METAL PENTAHALIDES. 
 
 49 
 
 This is produced, in an impure state as a black liquid solidi- 
 fying at about 73, by treating caesium triiodide with hot 
 water, and also by treating solid iodine with a hot solution of 
 caesium iodide. Artificial mixtures of caesium triiodide and 
 iodine, representing compositions varying from CsI 4 to CsI 9 , 
 all melt at a uniform temperature of about 73. It is evident 
 from this that the composition of the black liquid cannot be 
 determined from its melting-point. 
 
 Csesium triiodide, which is readily soluble in alcohol, be- 
 comes much more soluble in that liquid in the presence of two 
 atoms of iodine to the molecule. A very concentrated solu- 
 tion of this kind gives crystals of the pentaiodide by cooling, 
 but a much better product is obtained by concentration over 
 sulphuric acid, using a slight excess of iodine to allow for loss 
 by volatilization. The crystals are well formed and have a 
 brilliant black color. They can be distinguished from crystals 
 of iodine, which may separate if too much of this substance 
 has been used, by their brittleness as well as their form. The 
 substance melts, not sharply, at 73. It loses iodine on expo- 
 sure about as rapidly as iodine itself volatilizes. It does not 
 contain water or alcohol. 
 
 Samples of the crystals quickly dried wiiji paper gave the 
 following results on analysis : 
 
 Caesium 
 Iodine 
 
 Made by 
 cooling. 
 
 15.20 
 
 By evaporation. 
 Separate products. 
 
 20.96 16.02 
 
 Calculated 
 for CsI 5 . 
 
 17.32 
 82.68 
 
 When a concentrated solution of caesium bromide is shaken 
 up with a large excess of bromine, there is no separation of 
 caesium tribromide, as is the case when the theoretical amount 
 of bromine is used. A large part of the caesium bromide 
 goes into solution in the liquid bromine and, on taking up a 
 sufficient quantity of caesium bromide, this solution becomes 
 lighter in color than pure bromine. 
 
 4 
 
50 ON THE ALKALI-METAL 
 
 A solution of caesium bromide in bromine, made in the 
 manner above indicated, was allowed to evaporate spontane- 
 ously at a temperature below 0. A dark red solid finally 
 separated, and it was prepared for analysis by pressing with 
 papers at the same low temperature. After the adhering 
 bromine had been removed, the substance gave off bromine- 
 vapor very rapidly. 
 
 Analysis g ave g^gg 
 
 Caesium 29.93 24.95 
 
 Bromine 75.05 
 
 The analysis corresponds with the formula CsBr 5 as well as 
 could be expected, considering the great instability of the 
 compound. 
 
 CsCLClJ. 
 
 This substance can be prepared by dissolving 40 g. of 
 caesium chloride in a mixture of 600 c. c. of water and 200 
 c. c. of concentrated hydrochloric acid, adding 30 g. of iodine 
 (one atom), and passing chlorine to saturation; meanwhile 
 keeping the solution warm enough to dissolve any of the 
 compound which separates in the form of a yellow precipitate, 
 and finally cooling to crystallization. The hydrochloric acid 
 is used to prevent the simultaneous deposition of an acid 
 caesium iodate. 
 
 Analysis gave *$%$?* 
 
 Caesium 32.44 33.09 
 
 Chlorine 34.79 35.32 
 
 Iodine 31.11 31.59 
 
 The crystals are of a pale orange color. They are in the 
 form of slender prisms, usually in parallel position forming 
 plate-like groups. The body is sparingly soluble in water and 
 can be recrystallized from it without much decomposition. It 
 is nearly permanent in the air. On heating, it is apparently 
 converted into CsCl.CH, for it melts like that substance at 
 238 (uncorr.) in the open capillary tube. 
 
PENTAHALIDES, 51 
 
 RbCLCZ J. 
 
 This body can be conveniently prepared by adding 40 g. 
 of iodine to a nearly saturated solution of 38 g. of rubidium 
 chloride and passing in an excess of chlorine. The solution 
 becomes warm from the reaction and, on cooling, large orange- 
 yellow plates are deposited. 
 
 Analysis gave 
 
 Eubidium .... 24.12 23.63 24.11 
 
 Chlorine .... 39.00 . . . 40.05 
 
 Iodine ..... 35.31 . . . 35.83 
 
 The compound is soluble in alcohol, but unaffected by ether. 
 When rapidly heated in an open capillary tube, it melts at 
 213 (uncorr.), undergoing some decomposition, and becomes 
 completely white at about 270. These numbers agree quite 
 closely with the melting and whitening points of RbCl.ClI, 
 so that it is evident that there is a loss of two atoms of chlo- 
 rine before much further decomposition takes place. In view 
 of this fact, it is remarkable that, when samples of RbCl.Cl 3 I 
 and RbCl.CU were exposed to the air side by side for three 
 months, the compound containing the greater amount of 
 chlorine was almost completely decomposed, while the other 
 remained nearly unchanged. It is therefore probable that 
 RbCl.Cl 3 I decomposes at ordinary temperatures by losing 
 C1 8 I as a whole, while by heating another decomposition 
 takes place. 
 
 This compound, first described by Filhol, has been prepared 
 for the sake of studying its crystalline form. It is easily made 
 by the method which has been given for the corresponding 
 rubidium compound. The crystals obtained by cooling are 
 in the form of very slender needles, but, by evaporating the 
 mother-liquor from these at ordinary temperature, thicker 
 prisms suitable for measurement can be obtained. 
 
52 ON THE ALKALI-METAL 
 
 Analysis gave 
 
 Potassium 11.98 12.66 
 
 Chlorine 45.31 46.10 
 
 Iodine 42.50 41.23 
 
 NaCl.CUZH t O. 
 
 To prepare this substance, sodium chloride and iodine, in 
 the calculated proportions, are mixed with an amount of water 
 insufficient to dissolve the sodium chloride even on heating, 
 chlorine is added to saturation at a gentle heat, and the liquid 
 is filtered while still warm. The solution, on cooling to a low 
 winter temperature, gives a crop of slender needles, but better 
 crystals are obtained by evaporation in a desiccator. Some 
 of the latter, quickly dried on paper, gave the following results 
 on analysis : 
 
 fp nnn A Calculated for 
 
 NaCl.Cl 3 L2H 2 O. 
 
 Sodium 7.17 7.01 
 
 Chlorine 42.92 43.29 
 
 Iodine 38.23 38.71 
 
 Water 12.84* 10.97 
 
 The water was determined by direct weighing in a calcium 
 chloride tube, the halogens being retained by an ignited mix- 
 ture of lead oxide and lead chromate. 
 
 The body is rapidly decomposed by exposure. It melts 
 gradually between 70 and 90 and becomes white at about 
 115. It is decomposed by strong alcohol and by ether. 
 
 LiCl.ClJ.4HiO. 
 
 This was made by adding 60 g. of iodine to a hot saturated 
 solution of 20 g. of lithium chloride in dilute hydrochloric 
 acid, saturating with chlorine and cooling. A large quantity 
 of long yellow needles was thus obtained. On evaporating 
 the mother-liquor in a desiccator, larger prisms were deposited. 
 
 * Determined in a separate sample. 
 
PENTAHALIDES. 
 
 53 
 
 Analysis gave 
 
 Lithium 2.03 2.16 
 
 Chlorine 39.96 39.94 
 
 Iodine 37.54 36.77 
 
 Water . 20.93 
 
 Calculated for 
 LiCl.Cl 3 I.4HjO. 
 
 2.01 
 
 40.80 
 36.49 
 20.68 
 
 On exposure to the air the substance quickly deliquesces, 
 forming a yellow liquid. This gradually loses its color, finally 
 leaving a solution of lithium chloride. The body melts at 
 70-80 and becomes white at about 180. The crystals of 
 this compound were not measured. 
 
 Crystallography. 
 
 The crystallization of CsI 5 is triclinic. By slow evaporation 
 of solution in a desiccator, crystals were obtained which were 
 about 10 mm. in diameter. Two crops were examined, in one of 
 which the habit shown in Fig. 1 prevailed, while in the second 
 the crystals were more highly modified, like Figs. 2 and 3. 
 
 The forms which were observed are : 
 
 HP 
 
 a, 100, i-i 
 
 c, 001, 
 
 m, 110, 1' 
 
 M, 1TO, / 
 
 d, on, 
 
 /, 041, 4^' 
 6, 021, 2-i 
 PJ 311, -3-3' 
 
 0, 311, 3-3' 
 x, 341, -4-|' 
 y, 341, 4-|' 
 
 The axial ratio is as follows : 
 
 : I : c = 0.9890 ; 1 : 0.42765 
 a = 96 56' ft = 89 55V 7 = 90 21 y 
 
54 
 
 ON THE ALKALI-METAL 
 
 The crystals gave good reflections of the signal on the goni- 
 ometer. In the following tables the measurements which 
 were chosen as fundamental are indicated by an asterisk : 
 
 a 
 m 
 a 
 c 
 e 
 a 
 a 
 a 
 a 
 a 
 
 AC, 
 AM, 
 A m, 
 A e, 
 Arf, 
 AM, 
 AC, 
 
 Ad, 
 A/?, 
 AO, 
 
 100 A 001 
 
 no A no 
 
 100 A 110 
 001 A 021 
 021 A Oil 
 100 A 110 
 100 A 021 
 100 A Oil 
 100 A 311 
 100 A 311 
 
 Measured. 
 = *90 2' 
 = *89 47' 
 - *44o 43' 
 
 - *43 26' 
 = *65 25' 
 = 45 4' 
 = 90 6' 
 = 89 57' 
 = 41 18' 
 = 41 31' 
 
 Calculated. 
 
 45 4' 
 90 16' 
 89 54' 
 41 19' 
 41 25' 
 
 c 
 c 
 e 
 e 
 d 
 d 
 P 
 / 
 
 X 
 
 y 
 
 A TO, 
 AM, 
 A TO, 
 AM, 
 A TO, 
 AM, 
 A TO, 
 AC/, 
 A TO, 
 AM, 
 
 001 
 001 
 021 
 021 
 Oil 
 Oil 
 311 
 041 
 341 
 B41 
 
 Measured. 
 A 110 = 85 8' 
 A 110 = 85 7' 
 A 110 = 65 9' 
 A HO = 65 8' 
 A 110 = 70 3' 
 A HO = 70 & 
 A 110 = 40 41' 
 A Oil = 32 3V 
 A 110 = 25 46' 
 A 110 = 25 56' 
 
 Calculated. 
 85 9' 
 85 7f 
 65 3' 
 65 18' 
 70 2' 
 70 4 X 
 40 41' 
 32 37' 
 25 44|' 
 25 52' 
 
 The form of CsCl.Cl 8 I is monoclinic. From a 
 number of crystallizations this salt was always ob- 
 tained in needles, sometimes over 20 mm, in length 
 and having the habit shown in Fig. 4. 
 
 The forms which were observed are : 
 
 b, 010, i-l 
 I, 210, t-2 
 
 p, 212, 1-2 
 q, 211, -2-2 
 
 s, 211, 2-2 
 d, 041, 4-T 
 
 The axial ratio is as follows : 
 
 a:b:c = 0.9423 : 1 : 0.4277, = 100 A 001 = 86 20' 
 
 Measured. Calculated. 
 I A I, 210 A 210 = *50 22' 
 I A p, 210 A 212 = *44 61' 
 p A s, 212 A 2fl = *106 35' 
 / A q, 210 A 211 = 27 3' 27 T 
 
 Measured. Calculated. 
 
 I A s, 210 A 2fl = 28 35' 28 34' 
 
 p A p, 212 A 212 = 32 59' 33 2' 
 
 p A b, 212 A 010 = 73 31' 73 29' 
 
 b A rf, 010 A 041 = 31 0' 30 21' 
 
 The crystallization of RbCLClJ is 
 monoclinic. This salt was crystallized a 
 great many times, and was always ob- 
 tained in plates, sometimes over 20 mm. 
 broad, but seldom 1 mm. thick. The 
 habit is shown in Fig. 5. 
 
 The forms which were observed are : 
 
PENTAHALIDES. 55 
 
 b, 010, i-i m, 110, 1 s, Til, 1 
 
 c, 001, /?, Ill, -1 
 
 The axial ratio is as follows : 
 
 a : b : c = 1.1390 : 1 : 1.975, ft = 100 A 001 = 67 6J> 
 
 Measured. Measured. Calculated. 
 
 c A m, 001 A 110 = *74 26' CAS, 001 A 111 = 82 12' 82 15' 
 
 CAP, 001 A 111 = *55 20' s A m, 111 A 110 = 23 20 7 23 19' 
 
 , 111 A 111 = *76 21' />A TO, 111 A 110 =190 5' 19 6' 
 
 With the polarizing microscope the plates show an extinc- 
 tion parallel to their diagonals. In convergent light nothing 
 of the ring system can be seen, but a dark bar crosses the 
 field in the direction of the symmetry plane, indicating that 
 the plane of the optical axes is the clino-pinacoid. 
 
 The crystalline habits and axial ratios of CsCl.Cl 3 I and 
 RbCl.Cl 3 I are wholly different, and all attempts to find any 
 similarity or mathematical relation between them has failed. 
 We have endeavored to detect any hidden relation that might 
 exist by examining separate crops of crystals, made from a 
 solution containing both salts. Each form alone and mixtures 
 of both were thus obtained, but no crystals of an intermediate 
 form could be produced. One unmixed crop, having the 
 form and angles of CsCl.Cl 3 I, contained about sixteen per 
 cent of RbCl.Clgl, while another, having the form and 
 angles of RbCLClat, contained about eleven per cent of 
 CsCl.Cl 3 I. These results show that isomorphous mixtures 
 can be obtained of either form, depending upon which salt 
 predominates, while the absence of any intermediate forms, 
 and the inability to detect any mathematical relation between 
 the two kinds of crystals, lead us to believe that the com- 
 pounds are dimorphous. 
 
 The form of KCl.CLJ is monoclinic. This salt was re- 
 peatedly made in fine needle-like crystals, too small to meas- 
 ure, by allowing a warm saturated solution to crystallize. 
 By slow evaporation in a desiccator, at ordinary temperatures, 
 stouter prismatic crystals, over 20 mm. long and 2 mm. in 
 
56 
 
 ON THE ALKALI-METAL 
 
 6. 
 
 diameter were obtained, having the habit shown 
 JJ_ in Fig. 6. These gave excellent reflections and 
 were measured without difficulty at winter temper- 
 ature. 
 
 The forms which were observed are : 
 
 a, 100, i-i 
 m, 110, / 
 
 n t 120, t-2 
 d, 023, f-r 
 
 The axial ratio is as follows : 
 a:b:c = 0.9268 : 1 : 0.44725, = 100 A 001 = 84 18' 
 
 Measured. 
 
 roA>n, 110AlTO = *8522' 
 d A d, 023 A 023 = *33 3' 
 
 Measured. Calculated, 
 a A d, 100 A 023 = *84 32' 
 n A n, 120 A 120 = 56 58' 66 56' 
 
 The positions and crystal symbols which have been adopted 
 for this and the corresponding caesium salt were chosen to 
 show a similarity in the axial ratios. Both salts are alike in 
 having a prismatic habit, but the forms which occur on each 
 are quite different. If it were not for bringing out this simi- 
 larity in axial ratios, the crystallography of both salts could 
 be simplified somewhat by giving to the dome d above the 
 simpler indices Oil, and by taking the prism and pyramids 
 of the caesium salt as belonging to the unit instead of to the 
 macrodiagonal series. 
 
 The anhydrous alkali-metal pentahalides do not form a 
 well-defined crystallographic series, yet there are relations 
 between three of them which seem to us to be more than 
 coincidences. The similarity is shown in the following 
 table: 
 
 CsCLClgl Monoclinic a : b : c = 0.9423 : 1 : 0.4277, ft = 86 20' 
 KC1.C1 3 I " a : b : c = 0.9268 : 1 : 0.44725, ft = 84 18' 
 
 T -nr 5 * : * : = - 9890 : * : 0.42765 
 Tnchmc = %0 ^ = 8 
 
PENTAHALIDES. 
 
 57 
 
 The crystallization of NaCl.Cl 3 I.2H 2 O is ortho- 
 rhombic. By slow evaporation of a solution in a 
 desiccator, crystals were formed over 10 mm. in 
 length, having the habit shown in Fig. 7. 
 
 The forms which were observed are : 
 
 b, 010, i-i 
 m, 110, / 
 
 A HI, 1 
 
 d, 021, 2-i 
 
 m 
 
 7. 
 
 771 
 
 The axial ratio is as follows : 
 
 a:b:c = 0.6745 : 1 : 0.5263 
 
 The crystals were measured at a temperature near C. 
 and gave excellent reflections. 
 
 Measured. Measured. Calculated. 
 
 m A m, 110 A 110 = *68 0' m A b, 110 A 010 = 56 (X 66 0' 
 m A p, 110 A 111 = *46 44' 6 A d, 010 A 021 = 43 2^ 43 32' 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1892. 
 
ON SOME ALKALINE IODATES.* 
 
 BY H. L. WHEELER. 
 
 WITH CKYSTALLOGKAPHIC NOTES. 
 BY S. L. PENFIELD. 
 
 WHILE work on the compounds of iodine trichloride with 
 alkaline chlorides f was in progress in this laboratory, it was 
 noticed in making K1C.C1J, RbCl.Cl 3 I, and CsCl.CU that 
 white crystals were often formed under certain conditions. 
 These compounds proved to be KC1.KIO 8 .HIO 3 , RbCl.HIO 8 , 
 and 2CsIO 8 .I 2 O 6 . Since they were not analogous, although 
 formed under similar conditions, and since the rubidium and 
 caesium salts have not been described, an investigation of them 
 was undertaken. Attempts to prepare these compounds by 
 other methods led to the discovery of several other iodates. 
 The new compounds that have been prepared are as follows : 
 
 KbI0 3 CsI0 3 
 
 KbI0 3 .HI0 3 2CsI0 3 .I 2 5 
 
 KbI0 3 .2HI0 3 2CsI0 3 J 2 5 .2HI0 3 
 
 KbCl.HIOs CsCl.HIOs 
 3KbC1.2HI0 3 
 
 The compound which separated from the solution of the 
 potassium pentahalide has already been described, but since this 
 is a new method of preparation, and since there are conflicting 
 statements concerning its state of hydration, it has been re- 
 investigated. 
 
 The results of the investigation of the rubidium salts show 
 that the normal iodate is the only one of the series that can be 
 recrystallized unaltered from an aqueous solution. In the 
 
 * Amer. Jour. Sci., xliv, August, 1892. t Ibid., p. 42. 
 
ON SOME ALKALINE IODATES. 
 
 59 
 
 case of the csesium compounds, the normal iodate and the salt 
 2CsIO 3 .I 2 O5 are not decomposed by water. The other caesium 
 iodates give 2CsIO 3 .I 2 O 5 when recrystallized from water, and 
 not the normal iodate, thus showing an interesting difference 
 between the rubidium and caesium compounds. 
 
 It is the tendency of the acid rubidium iodates to separate 
 in a higher state of hydration than the corresponding csesium 
 compounds. 
 
 It is also an interesting fact that the formation of the com- 
 pounds of normal chloride and iodic acid was not observed on 
 mixing the constituents. In the case of rubidium, products 
 were obtained which proved to be RbIO 3 , RbIO 3 .HIO 3 , or 
 RbIO 3 .2HIO 3 , according to the concentration of the solutions 
 and the excess of RbCl or HIO 3 . On the other hand, by add- 
 ing hydrochloric acid to a solution of rubidium iodate, if the 
 acid is dilute RbIO 3 .2HIO 3 is formed, while if concentrated the 
 iodate is completely decomposed. Similar experiments, under- 
 taken with caesium chloride and iodic acid, did not give the 
 peculiar double compound CsCl.HIO 3 , but resulted in each 
 case in the formation of 2CsIO 3 .I 2 O 5 . 
 
 Method of Analysis. After the substances were prepared 
 for analysis as described in detail beyond, the halogens were 
 determined by first reducing the solution of iodate with sul- 
 phur dioxide, then precipitating with silver nitrate in the 
 presence of nitric acid. This precipitate was then heated in 
 a stream of chlorine, thus combining the test for chlorine and 
 its determination in one operation. In the filtrate from the 
 silver precipitate, the alkali metal was determined as sulphate 
 after the removal of the excess of silver by means of hydrogen 
 sulphide. Oxygen was determined in a separate portion by 
 precipitation with silver sulphate, drying the precipitate at 
 100, and then determining the loss on ignition. Duplicate 
 halogen determinations were then made in this residue. In 
 the case of the compounds containing the group I 2 O 3 , where 
 an error would be introduced if the oxygen was determined 
 in this manner, the substance itself was ignited and the oxygen 
 calculated from the loss. The presence of water in these com- 
 
60 ON SOME ALKALINE IODATES. 
 
 pounds was determined by directly weighing it in a calcium- 
 chloride tube, the substance being ignited in a combustion 
 tube containing a mixture of lead chromate and lead oxide. 
 
 Normal Rubidium lodate, RHO*. This compound was 
 made by adding one molecule of iodine pentoxide in either 
 strong or dilute aqueous solution, to a solution of one mole- 
 cule of rubidium carbonate. If the solutions are strong the 
 iodate separates as a sandy precipitate, but if they are hot and 
 dilute it separates on cooling in small grains or as a crystalline 
 crust. At 23, 100 parts of water dissolve 2.1 parts of the 
 salt. The compound, after filtering on the pump, washing 
 with a little water and drying on paper, gave the following 
 results on analysis : 
 
 Found. Calculated for RbIO s . 
 
 Eubidium 32.17 32.82 
 
 Iodine 48.50 48.75 
 
 Oxygen 20.59 18.43 
 
 The salt decrepitates strongly when heated, then melts, gives 
 off oxygen but no iodine, and the residue is rubidium iodide. 
 Hydrochloric acid readily dissolves it in the cold to a faint 
 yellow-colored solution which increases in color on standing. 
 On warming, chlorine is evolved and the solution turns bright 
 yellow from the formation of iodine trichloride. If boiled 
 with strong hydrochloric acid, RbCl.ClI * is formed, which 
 separates on cooling. 
 
 The formation of normal rubidium iodate was also observed 
 when a hot dilute aqueous solution of iodine trichloride was 
 treated with rubidium carbonate. The compound thus ob- 
 tained gave 48.43 per cent of iodine on analysis. It was 
 formed also by dissolving the acid iodate in a strong, hot solu- 
 tion of rubidium chloride, and allowing the mixture to crys- 
 tallize. This was identified by a rubidium determination 
 which gave 32.58 per cent. In general, the iodates of rubid- 
 ium all give this body when they are dissolved in hot water 
 and the solutions left to crystallize. The products obtained 
 
 * Amer. Jour. Sci., xliii, 475. 
 
ON SOME ALKALINE IODATES. 
 
 61 
 
 in this manner decrepitated on heating and did not give off 
 iodine. A rubidium determination in the substance obtained 
 from RbCl.HIOs gave 32.76 per cent; from 3RbC1.2HIO 8 , 
 32.22 per cent. 
 
 Acid Rubidium lodate, RbIO z .HIO y This was obtained by 
 mixing warm solutions of one molecule of iodine pentoxide 
 and two molecules of rubidium chloride. The compound 
 generally separates on cooling as a heavy crystalline powder. 
 It is difficultly soluble in cold water. Hot water dissolves it 
 more readily, and on cooling the normal iodate separates. It 
 is insoluble in alcohol. The crystals were filtered on a pump 
 and washed with a little cold water and then pressed on paper. 
 An analysis of these dried at 100 gave the following results, 
 the oxygen being determined by difference. 
 
 Found. 
 
 Rubidium 20.13 
 
 Iodine 58.12 
 
 Oxygen 21.46 
 
 Hydrogen 0.29 
 
 Calculated for RbI0 3 .HI0 3 . 
 19.58 
 
 58.19 
 
 21.99 
 
 0.23 
 
 The reaction which takes place in the preparation of this 
 compound is probably according to the following equation: 
 
 RbCl + 2HIO 3 = RbI0 8 .HIO a + HC1 
 
 The hydrochloric acid thus liberated reacts on a part of the 
 iodic acid, chlorine is evolved, and the solution becomes 
 yellow. When heated it does not decrepitate, but melts to 
 a yellow mass, gives off water, then iodine, and finally froths 
 with the evolution of oxygen. The residue consists of rubid- 
 ium iodide. 
 
 Diacid Rubidium lodate, RbI0^.2HIO z . For the prepa- 
 ration of this compound, 5 g. of RbIO 8 were dissolved in 
 50 c. c. of water with the aid of heat, then 13 g. of iodine 
 pentoxide in 50 c. c. of water were added, the mixture boiled 
 down to half its volume and allowed to cool. The body 
 separates as a heavy, crystalline powder. It is difficultly 
 soluble in cold water. When dissolved in hot water and 
 
62 ON SOME ALKALINE IODATES. 
 
 the solution left to crystallize, RbIO 3 separates. The product 
 obtained, as stated above, was separated from the mother-liquor 
 by filtering on the pump, washed with a little cold water and 
 dried at 100. 
 
 Found. 
 
 3 
 
 Rubidium .... 13.93 14.13 13.96 
 
 Iodine ..... 61.91 62.48 62.20 
 
 Oxygen ..... 23.74 . . . 23.51 
 
 Hydrogen .... 0.42 . . . 0.33 
 
 This compound does not lose water at 100. When heated 
 it does not decrepitate, but melts, gives off water, then iodine 
 and oxygen, leaving a residue of rubidium iodide. The com- 
 pound was also obtained by adding 10 c. c. of hydrochloric 
 acid sp. gr. 1.1 to 5 g. of RbIO 8 in 20 c. c. of water. The mix- 
 ture was warmed until all the RbIO 3 dissolved, when it gave 
 a faint yellow solution which slowly deepened in color. On 
 standing, a well-crystallized product of the compound under 
 consideration was obtained, containing 14.13 per cent of rubid- 
 ium and 62.19 per cent of iodine. 
 
 The addition of a saturated solution of rubidium chloride to 
 syrupy iodic acid produces a precipitate which dissolves again 
 in the excess of iodic acid. When more rubidium chloride 
 is added, the whole being kept over a lamp, a point is reached 
 where a precipitate begins to form in the hot solution. This 
 is the compound in question. It was identified by a rubidium 
 and an iodine determination. This gave 14.17 per cent of 
 rubidium and 61.83 per cent of iodine. 
 
 RbCl.HIOz. This salt can be made by simply allowing 
 a saturated solution of RbCl.Cl 3 I to stand for some hours, 
 when large colorless prisms form, attached to the plates of 
 RbCl.Cl 3 I. The solution, after removing the crystals, warm- 
 ing to dissolve the pentahalide and passing chlorine in 
 again, does not yield a further deposit of the substance. This 
 is explained by the fact that so much hydrochloric acid is 
 formed in the solution that the formation of this compound is 
 prevented. The crystals remain unaltered on exposure to the 
 
ON SOME ALKALINE IODATES. 63 
 
 air, but on treatment with cold water they are decomposed, 
 losing their lustre and becoming white. The solution has an 
 acid reaction towards litmus. The hot saturated solution of 
 this compound gives the normal iodate on cooling. The 
 material for analysis was mechanically separated from adher- 
 ing RbCl.Cl 3 I and dried in the air. 
 
 Vn , Calculated for 
 
 Found - EbCl.HI0 3 . 
 
 Kubidium . . . 28.88 . . . 28.78 
 
 Iodine .... 42.29 42.62 42.76 
 
 Chlorine .... 12.09 12.13 11.95 
 
 Oxygen .... 16.33 . . . 16.16 
 
 Hydrogen 0.26 0.33 
 
 This salt can also be prepared by adding a strong aqueous 
 solution of rubidium hydrate to a strong solution of iodine 
 trichloride in water. This gives at first a precipitate of the 
 compound 3RbC1.2HIO 3 , and the solution left at rest for a 
 few days gives the large well-developed crystals of RbCl. 
 HIO 3 unmixed with RbCl.Cl 8 I. These were identified by 
 their crystalline form. 
 
 On warming the crystals with hydrochloric acid, RbCl.Cl 3 I 
 is formed, probably according to the following equation : 
 
 RbCl.HIOs + 5HC1 = RbCLClJ + 3H 2 + Cl, 
 
 and the RbCl.Cl 3 I, on further heating, gives RbCl.ClI with 
 the liberation of chlorine. When the substance is heated it 
 melts, gives off water, chloride of iodine, and oxygen; the 
 residue consists of rubidium chloride and iodide. A deter- 
 mination of the halogens in this residue gave 3.52 per cent 
 of chlorine and 53.66 per cent of iodine. 
 
 3Kb 01. 2HIO*. This compound, which is analogous to the 
 sodium compound 3NaC1.2NaIO 3 .9H 8 O, described by Ram- 
 melsberg,* and also to the salt 3NaI.2NaIO 8 .19H 2 O, obtained 
 by Penny, f or 3NaI.2NaIO 8 .20H 2 O according to Marignac,t 
 except that it contains no water of crystallization, was pre- 
 
 * Pogg. Ann., xli, 648 ; cxv, 584. t Ann. Ch. Pharm., xxxvii, 202. 
 
 t Jabresb., 1857, 124 : Ann. Min., V, ix, 1. 
 
64 ON SOME ALKALINE IODATES. 
 
 pared by two methods. It was obtained by the addition of a 
 hot, strong, aqueous solution of rubidium hydroxide to a 
 strong solution of iodine trichloride, the latter being in excess. 
 The mixture was then filtered hot, and on cooling, a mass of 
 fine needles separated. The mother-liquor, on standing, yielded 
 the large crystals of RbCl.HIO 3 . The needles are stable in 
 the air and at 100. From the hot, saturated, aqueous solution 
 of the compound, the normal iodate separates on cooling. 
 
 The formation of this compound was also observed on 
 adding a strong solution of rubidium carbonate to a hot, 
 saturated solution of RbCl.Cl s I, the latter being in excess. 
 The colorless, slender, transparent needles, thus obtained, gener- 
 ally separate in groups radiating from a point on the surface 
 of the yellow crystals of RbCl.Cl s I. After separating the 
 colorless crystals mechanically from the pentahalide, they 
 were air-dried on paper and then analyzed, while the material 
 obtained according to the previous method was dried at 
 100. 
 
 From RbOH Prom Rb,CO 3 Calculated for 
 
 and IC1 3 . and RbCl.Cl 3 I 3RbC1.2HIO 3 . 
 
 Rubidium, 35.41 34.58 35.78 . . . 35.87 
 
 Iodine, 35.27 36.00 35.87 35.81 35.52 
 
 Chlorine, 14.99 14.82 15.26 15.16 14.90 
 
 Oxygen, . . . 13.15 . . . 13.64 13.43 
 
 Hydrogen, . . . 0.29 . . . 0.30 0.28 
 
 When heated, the substance does not decrepitate, but melts, 
 gives off chloride of iodine, and the residue consists of a 
 mixture of rubidium chloride and iodide. A sample of this 
 residue gave on analysis 9.68 per cent of chlorine and 38.91 
 per cent of iodine. 
 
 Normal Caesium lodate, OsIO s - This was prepared by add- 
 ing a moderately strong aqueous solution of iodic acid to a 
 strong solution of caesium carbonate, care being taken to have 
 the carbonate in excess. When all the iodic acid had been 
 added, the solution was boiled. On cooling, a crystalline 
 mass separated, consisting apparently of small cubes. At 24, 
 100 parts water dissolve 2.6 parts of the salt. It is insoluble 
 
ON SOME ALKALINE IODATES. 65 
 
 in alcohol. The body was prepared for analysis by filtering 
 on the pump, washing with cold water, and then pressing on 
 paper and drying .at 100. 
 
 Analysis gave Calculated for CsIO 3 . 
 
 Caesium .... 43.08 43.53 43.18 
 
 Iodine 40.84 . . . 41.23 
 
 Oxygen 15.74 . . . 15.59 
 
 This was also obtained, in attempts to prepare a ceesium salt 
 corresponding to 3RbC1.2HIO 3 , by adding caesium hydrate or 
 carbonate, in moderately strong aqueous solution, to a strong 
 solution of iodine trichloride in excess, when it at once sepa- 
 rated in the form of a white sandy precipitate, which under 
 the microscope was seen to consist of transparent grains of 
 indefinite form. Unless the iodine trichloride is nearly satu- 
 rated with the carbonate, CsCl.Cl 8 I or CsCl.ClI * is obtained, 
 mixed with the iodate. An iodine and oxygen determination 
 in the air-dried salt gave 40.55 and 40.83 per cent of iodine 
 and 15.67 per cent of oxygen. When this iodate is heated, it 
 does not give off iodine, but melts and evolves oxygen. The 
 residue is caesium iodide. 
 
 @CsIO s .I 2 5 . This substance can be prepared in pure 
 condition and in large quantity by mixing a moderately dilute, 
 aqueous solution of two molecules of caesium chloride with 
 one molecule of iodine pentoxide dissolved in a little water. 
 Any precipitate that may have been produced is dissolved by 
 the aid of heat and more water if necessary. On cooling, the 
 compound separates as a sandy powder. This can be washed 
 with water or recrystallized from hot water without decompo- 
 sition. It can also be recrystallized from dilute solutions of 
 iodic acid. At 21, 100 parts of water dissolve 2.5 parts of 
 this salt. It is insoluble in alcohol. The material for analy- 
 sis was air-dried after pressing on paper. 
 
 , Calculated for 
 
 Found. 2CsI0 8 .I 2 8 . 
 
 Caesium 27.93 28.00 
 
 Iodine 53.42 53.47 
 
 Oxygen 18.69 18.53 
 
 * Amer. Jour. Sci., HI, xliii, 17, and xliv, 42. 
 5 
 
66 ON SOME ALKALINE IODATES. 
 
 This compound invariably separates along with the crystals 
 of CsCl.Cl 8 I, when the latter is prepared in the absence of 
 hydrochloric acid, but the yield is not very large. It is thus 
 obtained in the form of small, rounded, white nodules, which, 
 on close inspection, are seen to occur in pairs, the two nodules 
 being on opposite sides of a thin layer of the pentahalide. 
 They were mechanically separated from the latter, no water 
 being used to wash the compound when prepared for analysis. 
 The f ollowing results are sufficient for its identification : 
 
 Caesium 29.11 
 
 Iodine 50.21 
 
 Oxygen 18.99 
 
 Chlorine 3.24 
 
 This compound was also obtained by the following methods. 
 By mixing 6 g. of CsIO 8 , 20 c. c. of water, and 10 c. c. of 
 HC1, sp. gr. 1.1. When the mixture was boiled, it became 
 yellow and chlorine was evolved, and when cooled the sub- 
 stance separated as a crystalline crust. It was identified by 
 a determination of caesium which gave 28.40 per cent. 
 
 The compounds 2CsIO 3 .l 2 O 5 .2HIO 3 and CsCl.HIO 3 give 
 this body when their hot saturated solutions are cooled. A 
 caesium determination in the products thus obtained gave 
 27.94 and 28.12 per cent respectively. 
 
 When this body is treated with hydrochloric acid, sp. gr. 
 1.1, the solution becomes yellow, evolves chlorine on warming, 
 and, when concentrated on the water bath, yields on cooling 
 well-crystallized CsCl.ClI. Analysis gave 50.68 per cent of 
 caesium chloride. (Calculated for CsCl.ClI, 50.90 per cent.) 
 
 When heated in a closed tube it gives no sign of water, 
 gives off iodine, then melts with the evolution of iodine and 
 oxygen. The residue consists of caesium iodide. 
 
 2CsIO s .I 2 Os.gITIO s .This body was obtained by adding 
 5 g. of 2CsIO 8 .I 2 O 6 to a boiling solution of 25 g. of iodine 
 pentoxide in sufficient water to form a syrup. Water was 
 then added, and the precipitate thus produced proved to be 
 the compound in question. Thus produced, it separates as 
 
ON SOME ALKALINE IODATES. 67 
 
 a finely divided amorphous precipitate, which can be dried in 
 the air or at 100 without losing water. It is difficultly solu- 
 ble in water and when crystallized from an aqueous solution 
 gives 2CsIO3.I 2 O 6 . An analysis of the substance dried at 
 100 gave: 
 
 u^mH Calculated for 
 
 2CsI0 3 .I 2 6 .2HI0 8 . 
 
 Caesium ..... 19.71 20.43 
 
 Iodine ...... 57.68 58.52 
 
 Oxygen ...... 20.41 20.89 
 
 Hydrogen ..... 0.12 0.16 
 
 Water determinations, in samples dried in the air on paper, 
 gave 1.45 and 1.38 per cent ; theory requires 1.44. 
 
 When the substance is heated it gives off water and iodine, 
 then oxygen, the residue consisting of caesium iodide. 
 
 Cs Cl.HIOy This was obtained, in an attempt to increase 
 the yield of 2CsIO 3 .I 2 O5, by adding a rather small quantity of 
 caesium carbonate to a hot, saturated solution of CsCl.Cl 3 I, 
 when, on cooling and allowing the mixture to stand, colorless, 
 flat, transparent prisms separated on the yellow crystals of 
 CsCl.Cl 3 I previously formed. These colorless prisms were 
 picked out from the solution, dried on paper and separated 
 mechanically, as far as possible, from any adhering CsCl.Cl 8 I. 
 These on analysis gave the following results : 
 
 Calculated for 
 CsCLHIO,. 
 
 Caesium ..... 38.09 . . . 38.60 
 
 Iodine ..... 36.08 36.29 36.86 
 
 Chlorine .... 11.69 11.82 10.31 
 
 Oxygen .... 13.85 . . . 13.94 
 
 Hydrogen .... 0.30 . . . 0.29 
 
 The crystals remain unaltered on exposure to dry air, but 
 on treating them with water they immediately become opaque. 
 On recrystallizing from water they give 2CsIO 8 .l2O 6 . When 
 the substance is heated, it gives off water and iodine chloride, 
 melts, and gives off oxygen, the residue consisting of chloride 
 and iodide of caesium. When it is warmed with hydrochloric 
 
68 ON SOME ALKALINE IODATES. 
 
 acid, it undergoes the same decomposition as the corresponding 
 rubidium compound. 
 
 KOl.KIO z .HIOs. This compound has previously been 
 prepared by treating KIO 3 with hydrochloric acid, or a solution 
 of iodine trichloride with potassium hydrate or carbonate. It 
 has been described by Serullas * and Rammelsberg f as anhy- 
 drous, and the formula 2KC1.2KIO 3 .I 2 O 5 was assigned to the 
 salt. Millon,J from his determination of potash in this salt, 
 concluded that the substance contained a molecule of water, 
 but he made no determination of it. Finally Marignac, who 
 examined it more carefully, made a determination of the water 
 by drying the substance at 100, then igniting it in a tube 
 with metallic copper and collecting and weighing the water 
 by means of a sulphuric acid tube. 
 
 The compound obtained from a solution of KC1.C1 3 I sepa- 
 rated in shining transparent prisms, stable in the air. It con- 
 tained water corresponding to the formula 2KC1.2KIO3.I 2 O 6 . 
 H 2 O or KC1.KIO 3 .HIO 3 . An analysis of the air-dried salt 
 gave the following results: 
 
 -p ni ,_^ Calculated for 
 
 KC1.KI0 3 .HI0 3 . 
 
 Potassium 16.94 16.83 16.82 
 
 Iodine 54.46 . . . 54.66 
 
 Chlorine 7.72 . . 7.64 
 
 Oxygen 20.66 
 
 Hydrogen ..... 0.20 . . . 0.22 
 
 This compound and the one obtained by Marignac are 
 therefore identical. 
 
 On ignition it gives off water, iodine chloride, and oxygen, 
 the residue consisting of potassium iodide and chloride. An 
 analysis of this residue gave 2.39 per cent chlorine and 70.87 
 per cent iodine. 
 
 The author takes occasion here to express his obligations to 
 Professor H. L. Wells for the use of the material in this in- 
 
 * Ann. Ch. Phys., II, xliii, 113. t Ibid., Ill, ix, 407. 
 
 t Pogg. Ann., xcvii. 
 
 Jahresb., 1856, 298. Ann. Min., V, ix, 1. 
 
ON SOME ALKALINE IODATES. 
 
 69 
 
 vestigation and for valuable suggestions; also to Professor 
 S. L. Penfield, who has kindly furnished the crystallographical 
 descriptions. 
 
 NOTES ON THE CRYSTALLINE FORM OF RbCl.HIO 8 AND 
 
 CsCl.HIO 8 . 
 
 The form of RbCl.HIO 3 is monoclinic. The 
 crystals are highly modified, doubly terminated 
 prisms, Fig. 1. The faces gave fair reflections, 
 and the measurements which were chosen as 
 fundamental are marked by an asterisk in the 
 table of angles. 
 
 The axial ratio and forms are as follows : 
 
 a : 3 : b = 0.9830 : 1 : 0.7577, = 100 A 001 = 87 56' 
 
 a, 100, i-l 
 
 b, 010, 'i-i 
 
 c, 001, 
 
 I, 320, i-I 
 TO, 110, I 
 n, 120, & 
 
 d, Oil, 1-i 
 
 e, 101, -1-i 
 /, 101, 1-i 
 
 9, 102, ft 
 a, 211, -2-2 
 ft HI, -1 
 
 q, 142, -24 
 s, 211, 2-2 
 u, 111, 1. 
 
 
 
 
 
 Measured. 
 
 Calcu 
 
 late. 
 
 a A 
 
 c, 
 
 100 
 
 A 001 
 
 = *87 56' 
 
 
 
 a A 
 
 e, 
 
 100 
 
 A 101 
 
 = *51 5' 
 
 
 
 C A 
 
 d.OOl 
 
 A Oil 
 
 = *37 8' 
 
 
 
 a A 
 
 I, 
 
 100 
 
 A 320 
 
 = 33 13' 
 
 33 
 
 13' 
 
 a A 
 
 M, 
 
 100 
 
 A 110 
 
 = 44 7' 
 
 44 
 
 29-1 
 
 a A 
 
 , 
 
 100 
 
 A 120 
 
 = 62 42' 
 
 63 
 
 1| 
 
 a A 
 
 0, 
 
 100 
 
 A 211 
 
 = 38 19' 
 
 38 
 
 28A 
 
 a A 
 
 P> 
 
 100 
 
 A 111 
 
 = 57 13' 
 
 57 14' 
 
 The form 
 the one crop 
 
 2. 
 
 ments could 
 mental are: 
 
 Measured. Calculated. 
 
 a A d, 100 A Oil =r 88 29' 88 21' 
 
 a A s, TOO A 211 = 38 32' 39 47' 
 
 a A , 100 A 111 = 59 57' 59 38' 
 
 e A p, 101 A 111 = 30 28' 30 31' 
 
 /AM, 101 A 111 -31 22^' 31 24' 
 
 p A g, 111 A 142 = 26 36' 26 30' 
 
 c A g, 001 A 102 = 21 17' 21 20' 
 
 c A /, 001 A 101 = 38 26' 38 23|' 
 
 CsCLHIOs 
 
 of CsCl.HIOg is monoclinic. The crystals, from 
 which was examined, were about 5 mm. in length 
 and had the habit shown in Fig. 2. 
 They were attached at one end, and 
 usually grew in radiating and diver- 
 gent groups. The faces were not very 
 perfect, and only approximate measure- 
 be made. Those which were chosen as funda- 
 
70 ON SOME ALKALINE IODATES. 
 
 TO A m, 110 AllO = 90 12' 7/1 A p, 110 A 221 = 24 37' a A P , 100 A 221 = 49 63 
 The axial ratio and forms are as follows : 
 
 d ~b : c = 0.9965 : 1 : 0.7698 = 100 A 001 = 89 53' 
 
 a,100,w m, 110,1 d,403, -fi s,403, fi p, 221, -2 
 
 c, 001,0 n, 130, i4 e, 203, -ft , 203, fi o, 263, -2-3 
 
 The pyramids p and o were frequently wanting. The 
 orthodomes d, e, s, and u were very constant in their develop- 
 ment and gave to the crystals an orthorhombic habit. Owing 
 to the curved and striated character of the faces, the symme- 
 try could not be satisfactorily determined by measurement, 
 but the optical properties showed that the crystals were truly 
 monoclinic. In polarized light the tables show an extinction 
 parallel to the ortho-axis, and in convergent light one of 
 the optical axes and the acute bisectrix can be seen near 
 the limits of the field. The plane of the optical axes is the 
 clino-pinacoid. 
 
 These two salts, although entirely different in crystalline 
 habit, are very similar in their axial ratios. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1892. 
 
ON A METHOD FOB THE QUANTITATIVE DETER- 
 MINATION OF CESIUM, AND THE PREPARATION 
 OF PURE OESIUM AND RUBIDIUM COMPOUNDS.* 
 
 BY H. L. WELLS. 
 
 SINCE no method has heretofore been devised for the accu- 
 rate quantitative determination of csesium in the presence of 
 both rubidium and potassium, some experiments have been 
 made in order to test the availability of the plumbic chloride, 
 described in a recent article, f for this purpose. The results 
 have not been as accurate as could be desired, but the method 
 will be useful until a better one is found. 
 
 The solubility of Cs 2 PbCl 6 in a hydrochloric acid solution 
 (fuming acid diluted with water 1:1), containing twice the 
 theoretical amount of lead chloride and saturated with chlorine, 
 was determined by making a precipitation of about 1 g. of 
 Cs 2 PbCl 6 under these conditions in 350 c. c. and determining the 
 csesium in the nitrate. The whole filtrate gave 0.0119 g. of 
 Cs 2 SO 4 , which corresponds to a solubility of 0.000068 g. of 
 Cs 2 PbCl 6 in 1 c. c. A similar experiment in which concentrated 
 hydrochloric acid was used, and also a larger excess of lead 
 chloride, gave a solubility of 0.00049 g. of Cs 2 PbCl 6 in 1 c. c. 
 It has been shown in the article referred to that the solubility 
 of Rb 2 PbCl 6 is 0.003 g. in 1 c. c. under similar conditions. 
 
 Some actual determinations of caesium were made as follows : 
 Known quantities of Cs 2 PbCl 6 and about an equal weight of 
 PbCl 2 were dissolved in hot HC1 (1:1). Chlorine was passed 
 into the solutions until they became cold, and, after standing 
 about three hours, the precipitates were collected in porcelain 
 Gooch crucibles and washed with hydrochloric acid containing 
 chlorine. The precipitates were decomposed with hot water, 
 
 * Amer. Jour. Sci., xlvi, September, 1893. t Ibid., p. 180. 
 
72 
 
 QUANTITATIVE DETERMINATION 
 
 and the csesium in the resulting solutions was determined as 
 sulphate. In one case a comparatively large amount of potas- 
 sium chloride was present. The details are as follows : 
 
 A 
 B 
 C 
 
 taken. 
 
 .1674 
 .1592 
 .1280 
 
 KCl 
 taken. 
 
 0.5 
 
 Volume 
 
 Cs 2 S0 4 
 
 Deficiency 
 
 1HC1. 
 
 found. 
 
 as Cs 2 SO 
 
 c. c. 
 
 35 
 
 .0856 
 
 .0026 
 
 35 
 
 .0807 
 
 .0031 
 
 35 
 
 .0638 
 
 .0035 
 
 These results indicate greater errors than were expected from 
 the previous solubility determinations. It is suspected that a 
 little of the precipitate was dissolved by washing, and the use 
 of hydrochloric acid containing lead chloride as well as chlo- 
 rine would probably diminish the error. The last experiment 
 shows that the presence of a considerable amount of potassium 
 has no influence upon the result. 
 
 The determination of csesium by this method can be simpli- 
 fied by weighing the precipitated caesium-plumbic chloride 
 directly. The salt is perfectly stable at 100. The following 
 table gives the details of a number of determinations made in 
 this way. The precipitates were all thoroughly washed with 
 hydrochloric acid containing chlorine and dried on an asbestos 
 filter at 100. 
 
 taken. 
 
 PbCl 2 
 taken. 
 
 KCl 
 
 taken. 
 
 Volume 
 HC1. 
 
 Cs 2 PbC! 6 
 found. 
 
 
 
 g- 
 
 A 
 
 0.2761 
 
 0.25 
 
 B 
 
 0.0878 
 
 1.0 
 
 C 
 
 0.1202 
 
 1.0 
 
 D 
 
 0.7558 
 
 0.1 
 
 E 
 
 0.2483 
 
 0.1 
 
 0.5 
 
 c.c. 
 
 * 
 
 g- 
 
 28 1 : 1 
 
 0.2650 
 
 0.0111 
 
 52 1 : 1 
 
 0.0833 
 
 0.0035 
 
 52 1 : 1 
 
 0.1071 
 
 0.0131 
 
 28 cone. 
 
 0.7369 
 
 0.0189 
 
 20 cone. 0.2359 0.0124 
 
 The results show considerable losses in csesium, which appar- 
 ently do not entirely depend upon the volume in which the 
 precipitation is made. It is believed that the losses occur 
 chiefly in washing, for large quantities usually show a larger 
 total loss than small ones. 
 
OF CAESIUM. 73 
 
 When caesium and rubidium are together, the precipitation 
 of caesium plumbic chloride is accompanied by a partial pre- 
 cipitation of the rubidium, unless the quantity of the latter is 
 small. It is possible, however, to make an indirect determina- 
 tion of the caesium in such a precipitate by weighing it and 
 afterwards determining the weight of the caesium and rubidium 
 sulphates. Two experiments have been made on this plan, 
 where not only rubidium, but also potassium, sodium, and 
 lithium were present. 
 
 A B 
 
 g. g. 
 
 Cs 2 PbCl 6 taken, 0.3561 0.1545 
 
 Rb 2 PbCl 6 taken, 0.2845 0.4101 
 
 To each of these were added about 0.15 g. each of potassium 
 and sodium chlorides, 0.25 g. of lithium carbonate, and 0.1 g. of 
 lead chloride. The substances were dissolved by boiling with 
 dilute hydrochloric acid, about an equal volume of concen- 
 trated acid was added, and chlorine was passed until the 
 solutions became cold. 
 
 A B 
 
 c. c. c. c. 
 
 Volume of solution, 30 50 
 
 After standing several hours, the precipitates were collected 
 on asbestos filters in porcelain Gooch crucibles, washed with 
 dilute hydrochloric acid saturated with chlorine, dried at 100 
 and weighed. 
 
 A B 
 
 Cs 2 PbCl 6 and Eb 2 PbCl 6 found, 0.5621 0.4538 
 
 The precipitates were treated on the filters with hot water, 
 the resulting solutions were evaporated with sulphuric acid, 
 the lead sulphate was removed by filtration, the filtrates were 
 evaporated and finally ignited in an ammoniacal atmosphere, 
 and the mixed sulphates were weighed. 
 
 A B 
 
 Cs 2 S0 4 and Eb 2 S0 4 found, 0.2826 0.2164 
 For calculating the results, the following formulae were used: 
 
74 QUANTITATIVE DETERMINATION 
 
 (P = weight of Cs,PbCl 6 + Eb 2 PbCl 6 ) 
 (S = weight of Cs 2 S0 4 + Kb 2 S0 4 ) 
 Weight of Cs = 5.095S - 2.301P 
 Weight of Eb = 2.006P - 3.801S 
 
 A B 
 
 Caesium taken 0.1381 0.0599 
 
 Caesium found 0.1464 0.0584 
 
 Error in caesium 0.0083+ 0.0015 
 
 Eubidium taken 0.0823 0.1186 
 
 Eubidium precipitated . . 0.0534 0.0876 
 
 The results show that approximate determinations of caesium 
 can be made by this method when all the alkali-metals are 
 present. The process leaves a part of the rubidium with the 
 potassium, and these two metals can be precipitated as platinic 
 chlorides and their amounts determined indirectly. 
 
 The method which has been described is useful for the ex- 
 traction of caesium and rubidium from their natural sources. 
 The following method of procedure may be suggested, sup- 
 posing all the alkali-metals to be present as chlorides in a 
 concentrated aqueous solution: 
 
 At least an equal volume of concentrated hydrochloric acid 
 is added, and any precipitated sodium and potassium chlorides 
 are removed. The solution is diluted somewhat, to avoid a 
 subsequent precipitation of these chlorides, a solution of lead 
 chloride, made by boiling lead oxide with a large excess of 
 hydrochloric acid, is gradually added, while chlorine is passed 
 into the solution until it is cold and until fresh additions of 
 lead chloride fail to produce a yellow precipitate. According 
 to my solubility determinations, this precipitation leaves less 
 than 1 g. of rubidium and a much smaller quantity of caesium in 
 each liter of the solution. The precipitate is usually almost 
 free from potassium. To ensure the complete purification of 
 the caesium and rubidium, the precipitate is washed with 
 hydrochloric acid containing chlorine and lead chloride, then 
 it is treated repeatedly with small quantities of boiling water 
 until completely decomposed, and the resulting solution is 
 
OF CAESIUM. 75 
 
 subjected to a repetition of the foregoing process. The mixed 
 plumbic salts are decomposed with hot water, and the resulting 
 filtered solution is evaporated to dryness to remove hydrochloric 
 acid. The residue is dissolved in hot water,* the lead is precip- 
 itated by the addition of a slight excess of ammonium sulphide, 
 and the precipitate is removed by filtration. The solution is 
 evaporated to dryness, and the residue consists of caesium and 
 rubidium chlorides and some ammonium chloride. 
 
 The following directions for the separation and purification 
 of the caesium and rubidium do not involve any new methods, 
 but the course of procedure has been arrived at after a consid- 
 erable amount of experience, and it may be of use to others. 
 It is assumed that rubidium is more abundant than caesium 
 in the mixture. If caesium predominated, it would be more 
 advantageous to extract that metal first by an obvious modi- 
 fication of the process. 
 
 The mixed chlorides of rubidium and caesium are dissolved 
 in at least five parts of concentrated nitric acid, and the solution 
 is evaporated to dryness and heated until the excess of nitric 
 acid is removed. The residue is dissolved in a small amount 
 of water, and as much oxalic acid as corresponds to twice the 
 weight of the original chlorides is added. The whole is evap- 
 orated to dryness, and the residue is ignited in platinum until 
 the oxalates are completely converted into carbonates. f The 
 carbonates are dissolved in water, the solution is filtered and 
 exactly neutralized with a measured solution of tartaric acid, 
 as much more tartaric acid as has been used for the neutraliza- 
 tion is added, and the solution is evaporated until it becomes 
 saturated while hot. The solution on cooling deposits acid 
 rubidium tartrate, which is washed with a small quantity 
 of water and is recrystallized two or three times from a hot 
 saturated solution, in the same way, until it gives no caesium 
 
 * No part of this residue should be thrown away on the assumption that it 
 is lead chloride, for the salt CsPb 2 Cl 6 is difficultly soluble and resembles 
 PbCl 2 . 
 
 t This method of converting alkaline chlorides into carbonates is due to 
 J. L. Smith, Amer. Jour. ScL, II, xvi, 373. 
 
76 DETERMINATION OF CESIUM. 
 
 spectrum.* The united mother-liquors from the acid rubidium 
 tartrate are evaporated to dryness and ignited in platinum. 
 The resulting carbonates are converted into chlorides, and, to 
 a solution of these in a small volume of 1 : 1 hydrochloric 
 acid, a solution of antimony trichloride in the same acid is 
 added as long as a precipitate forms, f The precipitate is 
 collected on a filter and washed with hydrochloric acid. To re- 
 move traces of rubidium, the precipitate is thoroughly decom- 
 posed with successive, small quantities of hot water, then 
 hydrochloric acid and a little antimony trichloride are added 
 to the whole, in order to repeat the precipitation. The last 
 precipitate is washed with hydrochloric acid. It usually shows 
 no rubidium when tested with the spectroscope. The caesium 
 antimony chloride is decomposed with hot water, and hydrogen 
 sulphide is passed into the resulting solution. The nitrate 
 from the antimony sulphide gives, on evaporation, pure caesium 
 chloride. The filtrates from the antimony double salt are 
 freed from antimony, evaporated to dryness, and the mixture 
 of caesium and rubidium chlorides, which should be very small 
 in amount, is preserved for use in subsequent purifications. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1893. 
 
 * This method is due to O. D. Allen, Amer. Jour. Sci., II, xxxiv, 367. 
 t Method of Godeffroy, Berichte, vii, 375. 
 
ON SOME PECULIAR HALIDES OF POTASSIUM 
 AND LEAD * 
 
 BY H. L. WELLS. 
 
 IN a recent article I have described a series of double chlo- 
 rides of the type M 2 PbCl 6 , where M is NH 4 , K, Rb and Cs. It 
 has seemed desirable to extend the investigation by attempt- 
 ing to prepare bromides and iodides corresponding to these 
 salts. A thorough search has been made, using the metals 
 of the potassium group and sodium, with the result that no 
 double bromides or iodides containing extra halogen could be 
 prepared, except in the case of potassium. It is remarkable 
 that the potassium-lead bromide and iodide which have been 
 discovered do not correspond in composition to the chlorides. 
 The failure to prepare double salts of rubidium and caesium 
 corresponding to the new potassium compounds was unex- 
 pected, for, as a general rule, the insolubility and stability and 
 the consequent ease of preparation of such compounds become 
 greater from potassium towards caesium. The explanation of 
 the anomaly probably lies in the fact that extremely concen- 
 trated rubidium and caesium solutions containing a lead halide 
 and the corresponding halogen cannot be obtained, in the case 
 of the bromides and iodides, on account of the slight solubil- 
 ity of caesium triiodide, and of the double halides which are 
 formed with PbBr 2 and PbI 2 . 
 
 The compounds to be described probably, have the composi- 
 tion represented by the following formulae : 
 
 K 8 Pb 2 I 8 .4H 2 
 K 8 Pb 2 Br 8 .4H 2 O 
 
 * Amer. Jour. Sci., xlvi, September, 1893. 
 
78 SOME PECULIAR HALIDES OF 
 
 These formulae may be also written, 
 
 3KI.2PbI 2 .I.4H 2 O and 3KBr.2PbBr 8 .Br.4H 2 
 
 The composition of these salts is very remarkable, on account 
 of the small amount of the extra halogen that they contain. 
 They apparently do not correspond to any other chemical com- 
 pound that is known. 
 
 The Iodide, K s Pb 2 I^H^ 0. This salt forms brilliant, 
 black, prismatic crystals, sometimes a centimeter or two in 
 length and three or four millimeters in diameter. Although 
 the crystals have fine prismatic faces, they never appear to 
 have definite terminations. The ends usually appear fibrous, 
 as though made up of numerous small crystals in parallel posi- 
 tion. When the crystals are crushed on paper it is evident 
 that they enclose much mother-liquor. The salt is deposited 
 from nearly, or quite, saturated solutions of potassium iodide 
 containing lead iodide and iodine. It is deposited at ordinary 
 temperature, usually slowly, after the lapse of several hours or 
 even after several days. In preparing the compound, the lead 
 iodide and the iodine can be varied considerably, but it is 
 formed only in very concentrated potassium iodide solutions, 
 and it is difficult to obtain crops of it which are not evidently 
 contaminated with this salt in the form of crystals. The salt 
 is stable in the air, but it is instantly decomposed by water or 
 alcohol, so that it cannot be washed. 
 
 Six separate crops have been analyzed, and great care has 
 been used in selecting them and in drying -them on paper for 
 analysis. In two cases the product was rapidly and finely 
 pulverized during the drying operation, but without any effect 
 upon its composition. The results of the six analyses agree 
 with remarkable closeness, but in spite of this fact it must be 
 assumed, from considerations which will be given subsequently, 
 that all these products were seriously contaminated with potas- 
 sium iodide. The fibrous nature of the crystals, and the con- 
 centration of the mother-liquor, make the possibility of such a 
 contamination very evident, but the constancy of this contam- 
 ination, as indicated by the uniformity of the analyses, is very 
 
POTASSIUM AND LEAD. 79 
 
 remarkable in view of the fact that some of the products were 
 made at wide intervals of time, covering a period of about six 
 months, so that there were considerable variations in the 
 laboratory temperature. 
 
 The products were made under the following conditions : 
 
 KI. Phi* I. Volume, 
 
 g. g. g. c.c. 
 
 A ... 450 30 15 ? 
 
 B ... 425 30 50 450 
 
 C ... 445 40 70 470 
 
 D ... 445 40 100 470 
 
 E ... 445 40 150 460 
 
 F ... 200 15 15 200 
 
 They gave the following results on analysis : 
 
 A . . 
 
 B . . 
 
 C . . 
 
 D . . 
 
 E . . 
 
 F . . 
 
 In these analyses, and those which follow, water was deter- 
 mined by weighing it directly in a calcium-chloride tube. 
 The other determinations were made according to the methods 
 mentioned in the preceding article on the double salts of lead 
 tetrachloride. 
 
 The above analyses correspond closely to the formula 
 K 9 PbJi 9 .10H 2 O, but it will be shown beyond that the proba- 
 ble formula of the pure compound is K 8 Pb2l8.4H 4 O. This 
 requires K = 7.25, Pb - 25.56, I - 62.74, and H 2 O = 4.45. If 
 this is the true composition, it must be assumed that all of the 
 analyzed products were contaminated with about 16.5 per cent 
 of potassium iodide, and that an excess of water was present, 
 possibly on account of the hygroscopic properties of that salt. 
 
 It is to be noticed that the products were prepared under 
 great variations in the amount of iodine present, and it can 
 be safely assumed, from the care with which the products 
 
 K. 
 
 Pb. 
 
 i. 
 
 H 2 0. 
 
 
 9.31 
 
 22.03 
 
 64.00 
 
 4.69 = 
 
 100.03 
 
 9.25 
 
 22.30 
 
 . . . 
 
 4.81 
 
 . . . 
 
 9.07 
 
 22.03 
 
 63.98 
 
 4.89 = 
 
 99.97 
 
 9.21 
 
 21.98 
 
 64.09 
 
 4.71 = 
 
 99.99 
 
 9.20 
 
 22.13 
 
 64.17 
 
 
 
 9.27 
 
 22.02 
 
 63.84 
 
 
 
80 SOME PECULIAR HALIDES OF 
 
 were examined, that they were not contaminated with the salt 
 KPbI 3 .2H 2 O nor any similar compound. The amount of lead 
 iodide in the solutions was comparatively small, and a large 
 part of it was used in forming the salt under consideration, so 
 that any contamination must have been chiefly potassium iodide. 
 It is therefore evident, since the salt is not decomposed on 
 exposure, and since the analyses show a constant amount of 
 extra iodine in spite of the variations of this ingredient in the 
 solutions, that the analyses must show the true ratio between 
 the lead iodide and the extra iodine in the pure compound. 
 This ratio is 2PbI 2 : 1 in both K 9 Pb J 19 and K 3 Pb 2 I 8 . 
 
 The Bromide, K^Pl 2 Br^H z 0. This compound forms 
 dark brown, prismatic crystals, which are solid and definitely 
 terminated, so that they do not have the tendency to hold 
 inclosed mother-liquor which the iodide has. The salt is easily 
 prepared and it crystallizes well, but it is extremely unstable. 
 When exposed to the air, it begins to whiten almost instantly, 
 giving off bromine. It is stable, however, in air containing a 
 considerable amount of bromine vapor, so that it can be dried 
 by pressing on paper in such an atmosphere. It is sufficiently 
 stable, when corked up in a weighing-tube, to be rapidly 
 weighed in a cold room without serious decomposition. 
 
 Three crops of the double bromide were analyzed. A and 
 B were made, in each case, by adding 20 c. c. of bromine to 
 400 c. c. of a cold solution which was saturated with potas- 
 sium bromide and lead bromide, and allowing the mixture to 
 stand over night. C was made like the other crops, except 
 that 30 c. c. of bromine were used. 
 
 Found. Calculated for 
 
 Potassium . 
 Lead . 
 
 A. 
 
 . . 10.33 
 . . 32.05 
 
 B. 
 
 10.41 
 31.90 
 
 c. 
 
 10.24 
 32.49 
 
 K 3 Pb 2 Br 8 .4H 8 0. 
 
 9.43 
 33.30 
 
 Bromine 
 Water . . 
 
 . . 51.96 
 
 52.15 
 5.59 
 
 52.05 
 5.28 
 
 51.48 
 5.79 
 
 100.05 100.06 100.00 
 
 The analyses agree with the formula as well as could be 
 expected, considering the instability of the compound. The 
 
POTASSIUM AND LEAD. 
 
 81 
 
 analyses show, almost exactly, one atom of extra bromine for 
 two atoms of lead, so that the compound is closely related to 
 the iodide, if not exactly analogous to it. 
 
 The satisfactory crystals of the bromide, and the stability of 
 the iodide, suggested the possibility that, if the two compounds 
 were really analogous, as suspected, isomorphous mixtures of 
 the two could be made which would retain the desirable quali- 
 ties of both, so as to be solidly crystallized and stable enough 
 to be accurately analyzed. Experiments showed that isomor- 
 phous mixtures could be readily obtained which crystallized 
 satisfactorily, and it was found that even small amounts of 
 iodine had the effect of greatly increasing the stability of the 
 compound. It was noticed that when a product was made 
 from a solution containing free bromine and iodine in nearly 
 atomic proportions (BrI), an almost perfectly stable, bright 
 red salt was obtained. The color of this salt is far from being 
 intermediate between that of the black iodide and the dark 
 brown bromide, but, since the analyzed products contain 
 about 23 atoms of bromine to one of iodine, it does not seem 
 probable that any definite relation between the two halogens 
 exists. It is remarkable that such a small proportion of 
 iodine should have so great an influence upon the color and 
 stability of the product, but it is to be noticed that only one- 
 eighth of the halogens in these compounds is in excess, so 
 that, if all the iodine is in this condition, it amounts to about 
 one-third of this excess. 
 
 The crops A and B had a dark bronze color. They were 
 successive crops, made by adding bromine to a strong solution 
 of potassium iodide containing lead iodide. The exact con- 
 ditions are unknown, but it is probable that insufficient 
 bromine was used to set free all the iodine which the solution 
 contained. These products were apparently as stable as the 
 iodide. 
 
 C and D were successive crops, made by continuing the addi- 
 tion of bromine to a somewhat similar solution until a change 
 of color showed that the free iodine had been converted into 
 BrI. These salts were red. An analysis of the mother-liquor 
 
 6 
 
82 SOME PECULIAR HALIDES OF 
 
 from D gave, KBr = 31.3, PbBr 2 = 1.8, Br = 6.7, I = 8.3, 
 H a O (difference) = 51.9. 
 
 E was made by adding 31 g. of bromine to 430 g. of the 
 above-mentioned analyzed solution. This crop was also red, 
 but it was not quite as bright in color and not as stable as 
 the others. On continuing the addition of bromine, still less 
 stable crops were obtained, which approached the pure bromide 
 in color. These were not analyzed. 
 
 The analyses of the five crops are as follows : 
 
 A . 
 B . 
 
 K. 
 
 . . 9.41 
 . . 9.24 
 
 Pb. 
 
 31.57 
 31.55 
 
 Br. 
 41.40 
 
 39.27 
 
 i. 
 12.06 
 14.57 
 
 H a O. 
 
 5.09 = 
 
 99.53 
 
 C . 
 
 D . 
 
 E , 
 
 . . 9.90 
 . . 9.99 
 . 10.24 
 
 32.88 
 32.74 
 32.26 
 
 48.66 
 48.70 
 49.97 
 
 3.40 
 3.30 
 2.07 
 
 5.24 = 
 5.02 = 
 
 100.08 
 99.75 
 
 The ratios calculated from the above analyses are as follows : 
 
 K : Pb : Br + I : H 2 0. 
 
 A 1.57 1. 3.99 1.83 
 
 B 1.55 1. 3.99 
 
 C 1.59 1. 4.00 1.82 
 
 D 1.61 1. 4.00 1.76 
 
 E 1.68 1. 4.16 
 
 The ratio required for the formula K 3 Pb 2 (Br,I) 8 .4H 2 is 
 
 K : Pb : Br + 1 H,O. 
 
 1.50 1. 4. 2. 
 
 The analyses agree well with this formula, except that the 
 water is somewhat low. Although 3J molecules of water 
 would correspond more closely to these analyses than 4, the 
 latter number is considered more probable, on account of the 
 fact that the analyses of the iodide show some excess over 
 four molecules. 
 
 It is to be seen that these mixed salts correspond in com- 
 position to the bromide. The analogous mode of formation 
 of the iodide, the identical relation of the lead to the extra 
 halogen in the iodide and the other products, as well as the 
 
POTASSIUM AND LEAD. 83 
 
 existence of these mixed salts, make it appear certain that 
 the analyzed iodide was invariably impure, and that the pure 
 compound should be considered as analogous to the other 
 salts. This view has been confirmed by a crystallographic 
 examination of the iodide and the red bromo-iodide, which 
 Prof. S. L. Penfield has kindly undertaken. He has found 
 that both these salts crystallize in prisms of the tetragonal 
 system. Unfortunately the crystals of the iodide were with- 
 out terminations, so that a more detailed comparison of the 
 two salts could not be made. 
 
 The nature of these peculiar salts is not clear. If they 
 are, strictly, hydrous " double salts," such higher halides as 
 Pb 2 Is or K 8 I 4 must be assumed. If they are formed from such 
 compounds as PbI 4 or KI 8 , they must be considered as hydrous 
 triple salts. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1893. 
 
ON THALLIUM TRIIODIDE AND ITS RELATION 
 TO THE ALKALI-METAL TRIIODIDES.* 
 
 BY H. L. WELLS AND S. L. PENFIELD. 
 
 THE well-known resemblance between the thallous salts and 
 many of the corresponding alkali-metal salts has led us 
 to prepare thallium triiodide and to compare its crystalline 
 form with that of the alkali-metal triiodides.f As a result, 
 it has been found that T1I 3 agrees, with remarkable closeness, 
 in form with RbI 3 and CsI 3 , and thus a case of isomorphism 
 is established between the higher iodides of thallium and the 
 alkali-metals. 
 
 This isomorphism is of special interest because our study 
 of the trihalogen compounds of csesium has led us to the con- 
 clusion that these have the structure of double salts. We 
 consider the evidence of this double-salt structure as very 
 strong, and since it seems necessary to infer that isomorphism 
 indicates the same arrangement of the atoms, we are obliged, 
 in spite of the apparent trivalence of thallium in thallic com- 
 pounds, to conclude that T1I 3 is also a double salt, to which 
 the formula T1I.I 2 should be given. It is not safe to assert 
 at present that all thallic salts must be similarly constituted, 
 for it is possible that thallium triiodide is not a true thallic 
 compound at all, and that thallic sulphate, nitrate, etc., have 
 an entirely different kind of structure. If it is granted that 
 thallium triiodide is a double salt, it seems probable that many 
 other compounds, which are considered as showing higher 
 valence of elements, may, in reality, have the structure of 
 double salts or " addition products." 
 
 Thallium triiodide was first described by Nickles,J who pre- 
 pared it by evaporating an ethereal solution of thallous iodide 
 
 * Amer. Jour. Sci., xlvii, June, 1894. 
 
 t Ibid., in, xliii, 17 and 475. } J. Pharm. [4], 1, 25. 
 
ON THALLIUM TRI IODIDE. 
 
 85 
 
 and iodine. Nickls states that he did not obtain it in -a pure 
 condition, but that his product always contained an excess of 
 iodine. He described its crystalline form, and his results 
 will be mentioned beyond. 
 
 We have modified Nickles' method by using alcohol as a 
 solvent, and have encountered no difficulty in obtaining a 
 pure product. The amount of iodine used was slightly in 
 excess of the calculated quantity, and the solution, produced 
 after long digestion, was evaporated over sulphuric acid until 
 crystallization took place. The resulting crystals were fre- 
 quently of large size, perfectly black, with a magnificent lustre 
 which was slowly lost upon exposure. A sample of the salt, 
 simply pressed upon paper, gave the following results upon 
 analysis : 
 
 'alcula^d 
 
 Thallium . 34.22 34.87 
 
 Iodine . 
 
 64.80 
 
 65.13 
 
 An examination of the crystals has shown that they are 
 orthorhombic and isomorphous with the orthorhombic alkali- 
 metal trihalides. Moreover, all the forms which have been 
 observed have also been found on the alkali-metal salts, and 
 are as follows : 
 
 a, 100, irl 
 
 b, 010, i-i 
 
 1. 
 
 c, 001, o 
 
 ff , 012, 
 
 d, Oil, 1-T p, 111, 1 
 ?, 102, \-l 
 
 a. 
 
 a 
 
 The habit is shown in Figs. 1 and 2, the latter being 
 remarkably like that of CsI 8 , when this had been crystallized 
 
86 ON THALLIUM TRIIODIDE. 
 
 from alcohol. The measurements which were chosen as funda- 
 mental are d A d, Oil A Oil = 96 34' and e A e, 102 A 102 = 
 78 48', giving the axial ratio : 
 
 a:b:c = 0.6828 : 1 : 1.1217 
 
 The dome g was determined by the measurement g A #, 
 012 A 012 = 58 34', calculated 58 34', and the pyramid p by 
 its position in the zones a d and d e. 
 
 A description of this salt, including a figure, has been given 
 by Nickle"s. His salt, crystallized from ether, had the habit 
 shown in Fig. 3, the letters in brackets being those used by 
 him and the position being changed to correspond with the 
 orientation of the alkali-metal trihalides. He considered p as 
 a prism, t as a macropinacoid, and m and n as brachydomes. 
 No calculations are given, and only the following four 
 measurements : 
 
 Nickles. Measured. Calculated from author's measurement. 
 
 p A p = 100 15' 101 12' for e*e, 102 A 102 
 
 p A t = 39 22' 39 24' " e A c, 102 A 001 
 
 p*m= 61 59 3' A rf, 102 A Oil 
 
 n* t= 19 25' 20 30' " 013 A 001 
 
 The agreement between the measured and calculated angles 
 is not very close, but Nickles' measurements cannot be very 
 exact, for if we take p A = 39 22' and n A t=19 25' as 
 fundamental, we find by calculation ^>A^ = 10116' and 
 p A m = 57 55', which vary considerably from his measure- 
 ments. Nickles crystals differ from ours not only in habit but 
 in having the one-third brachydome w, 013, which has not 
 been observed either in the T1I 3 prepared from alcohol or on 
 any of the alkali-metal trihalides prepared by us. 
 
 The very close agreement between the forms of rubidium, 
 caesium, and thallium triiodides is to be seen from the follow- 
 ing table of axial ratios : 
 
 RbI 8 . . . a:b:c = 0.6858 : 1 : 1.1234 
 CsI 8 ...= 0.6824 : 1 : 1.1051 
 T1I 8 . . . " " " = 0.6828:1:1.1217 
 
ON THALLIUM TRI IODIDE. 87 
 
 Our previous observation, that the exchange of one metal 
 for another in the trihalogen compounds usually has little or 
 no effect upon the crystalline form, is strongly confirmed by 
 these ratios, and the remarkable agreement between the ru- 
 bidium triiodide and the thallium compound is very striking, 
 when the great difference between the atomic weights of the 
 two metals is considered. 
 
 It was hoped that a pentaiodide of thallium could be pre- 
 pared, in order that its form might be compared with that of 
 caesium pentaiodide, but, by the use of increasing proportions 
 of iodine with thallium triiodide in alcoholic solutions, no evi- 
 dence of the existence of such a compound could be obtained. 
 
 The remarkably close relations of thallium to the alkali- 
 metals, as far as the thallous compounds are concerned, and 
 the additional resemblance which has been pointed out in the 
 present communication, have led us to consider the possibility 
 that thallium has been wrongly placed in the periodic system 
 of the elements and that it really belongs to the alkali -metals. 
 There are two vacancies in Mendele*eff's table in the alkali- 
 metal group corresponding to atomic weights of about 170 and 
 220. One of these is smaller, the other larger than the 
 accepted atomic weight of thallium, so that, as far as these 
 numbers are concerned, thallium might be composed of two 
 alkali-metal elements. Although the probability that thal- 
 lium was composed of two elements seemed very slight from 
 other considerations, we have deemed it desirable to test the 
 question experimentally. 
 
 About 200 g. of thallium were converted into the nitrate, 
 and this was systematically fractionated by crystallization, 
 until about one-twentieth of the salt remained as a repeatedly 
 recrystallized portion, and about another twentieth was con- 
 tained in a final mother-liquor. From each of these two frac- 
 tions, thallous chloride was prepared by converting into 
 sulphate, precipitating impurities with hydrogen sulphide, and 
 finally precipitating thallous chloride by means of hydrochloric 
 acid. The preparations were carefully washed, dried at 100, 
 and the chlorine was determined as silver chloride in order to 
 
88 ON THALLIUM TRIIODIDE. 
 
 get the atomic weight of the metal in each fraction. The silver 
 chloride was weighed hi the Gooch crucible, a method which 
 can be most highly recommended for accurately weighing this 
 substance. The following results were obtained, the weights 
 being given as taken in air : 
 
 Crystallized End. Soluble End. 
 g- g- 
 
 T1C1 taken 3.9146 3.3415 
 
 AgCl obtained 2.8393 1.9968 
 
 Atomic weight of Tl (0 = 16) . 204.5 204.5 
 
 It was not expected that absolute accuracy in the atomic 
 weight of thallium would be attained, but since the same 
 method of purification and analysis was used in both cases, the 
 two results are comparable with each other, and their exact 
 agreement shows that the fractionation of the nitrate gives no 
 change in the atomic weight of thallium, and no evidence has 
 been obtained that thallium is not homogeneous. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 January, 1894. 
 
ON SOME COMPOUNDS CONTAINING LEAD AND 
 EXTRA IODINE.* 
 
 BY H. L. WELLS. 
 
 ABOUT two years ago, the writer described f the double salts 
 of lead tetrachloride, (NH 4 ) 2 PbCl 6 , K 2 PbCl 6 , Rb 2 PbCl 6 , and 
 Cs 2 PbCl, and upon attempting to prepare the corresponding 
 bromides and iodides, an entirely different kind of double salts 
 was discovered.:): These peculiar salts were K 8 Pb 2 Br 8 .4H 2 O 
 and K 3 Pb 2 l8.4H 2 O. They are remarkable in containing but a 
 single atom of extra halogen hi the formula as given above, 
 and they apparently correspond to no previously known com- 
 pound. I was unable to obtain, with the alkali metals, any 
 bromides or iodides corresponding to the chlorides, but it is 
 interesting to notice that Classen and Zahorski have obtained 
 such salts with quinoline, (C 9 H,NH) 2 Pb 6 Br 6 and (C 9 H 7 NH) 2 
 PbI 6 .|| 
 
 The isolation of lead tetrachloride by Friedrich,^[ and the 
 discovery of lead tetraacetate, Pb(CH 3 CO 2 ) 4 , by Hutchinson 
 and Pollard,** were very interesting additions to our knowledge 
 of the compounds of tetravalent lead. These articles appeared 
 almost simultaneously with that of Classen and Zahorski, which 
 has been referred to above, and with my own work mentioned 
 at the beginning of this article. 
 
 As a sequence to my former investigations, it has seemed 
 to be desirable to reinvestigate two previously described com- 
 pounds containing lead and extra iodine, because it seemed 
 
 * Amer. Jour. Sci., 1, July, 1895. 
 
 t Ibid., xlvi, 180, 1893. " t Ibid., 190, 1893. 
 
 Zeitschr. fur anorg. Chem., iv, 107, 1893. 
 
 || Classen and Zahorski gave a formula of different type, 6NH 4 C1.2PbCl 4 , 
 to the double ammonium chloride. It seems certain from analogy, from 
 Friedrich's results, and from my own work, that their product was contami- 
 nated with ammonium chloride. 
 
 IF Berichte, xxvi, 1434, 1893. ** Chem. Soc. Jour., Ixiii, 1136, 1893. 
 
90 ON SOME COMPOUNDS CONTAINING 
 
 possible that a further study of them might throw some light 
 upon the nature of the curious salt, K 3 Pb 2 I 8 .4H 2 O. 
 
 Johnson's Salt. By mixing a hot, concentrated alcoholic 
 solution of potassium triiodide with a saturated solution of 
 lead acetate in boiling alcohol, filtering off the small precipi- 
 tate thus produced, and cooling, G. S. Johnson * obtained a 
 crystalline substance to which he gave the formula, Pb 8 C 86 H 54 
 O 2 8K 6 Ii 7 . Concerning this he remarks, "The formation of a 
 rational formula has at present baffled all my endeavors." 
 
 Johnson also obtained the salt by recrystallization from alco- 
 hol and by evaporating the mother-liquor over sulphuric acid, 
 but there is no evidence that he analyzed more than one sam- 
 ple of it. He does not give the quantities used in making his 
 preparation. 
 
 I have made a large number of crops of the compound, all 
 of which agreed with Johnson's description in forming rectan- 
 gular crystals, of a black color, having a marked brassy lustre 
 upon four of the six faces, and occurring usually in intergrown 
 groups of nearly square, flat plates. In preparing these prod- 
 ucts the conditions were varied considerably. As a starting- 
 point 30 g. of potassium iodide and 50 g. of iodine were 
 invariably used. These amounts give a slight excess of iodine 
 over the proportion required for potassium triiodide. From 
 40 to 100 g. of crystallized lead acetate were used, and it was 
 found that beyond these limits the preparation was unsuccessful. 
 
 The solvent varied from absolute alcohol, diluted only with 
 the water of crystallization of the lead acetate, to alcohol 
 diluted with one-half its volume of water. Several crops were 
 prepared in the presence of glacial acetic acid, and a volume of 
 this amounting to T V of the total liquid (20 c. c.) was used with 
 success. The total volume of solvent varied from 200 to 
 500 c. c., the larger amounts being used when it was not ex- 
 pected to obtain the product by simple cooling. It was custom- 
 ary to dissolve the potassium iodide and iodine in about one-half 
 of the solvent to be used and the lead acetate in the remainder. 
 The solutions were sometimes mixed boiling hot, while at 
 
 * Chem. Soc. Jour., xxxiii, 189, 1878. 
 
LEAD AND EXTRA IODINE. 91 
 
 other times a lower temperature was employed. A precipitate, 
 evidently consisting chiefly of lead iodide, was always pro- 
 duced by mixing the two liquids, but its quantity was usually 
 small. The effect of the presence of iodine in preventing the 
 precipitation of lead iodide to a great extent is very remark- 
 able. The solutions were filtered, sometimes while hot, some- 
 tunes after a longer or shorter period. The products obtained 
 by cooling formed coherent crusts composed of very small, 
 intergrown crystals, while by evaporation over sulphuric acid 
 much larger isolated crystals, or groups of crystals, were 
 deposited. All the analyses given below were made upon 
 crops obtained by evaporation, except in one instance. Two 
 partial analyses of products made by cooling are not included 
 in the list, because the results varied rather widely from each 
 other and from the results obtained with the products of evapo- 
 rations. The omitted results differed still more from Johnson's 
 analysis than the others. Two or three successive crops were 
 often obtained by evaporating a single solution, and the twelve 
 products, analyses of which are given, represent six different 
 original solutions. The products were well crystallized and 
 most of them seemed entirely satisfactory in regard to purity. 
 They were all examined microscopically, and as far as could be 
 judged from the appearance of an opaque substance, no im- 
 purities were present. The samples for analysis were very 
 carefully pressed upon filter-paper in order to remove the 
 mother-liquor. The salt is practically stable in the air, so that 
 decomposition was not to be feared during the drying operation. 
 Lead and potassium were determined by dissolving the sub- 
 stance in dilute nitric acid, evaporating with sulphuric acid, 
 separating the lead sulphate by filtration, weighing it, and 
 determining potassium in the filtrate by weighing it as sul- 
 phate. Iodine was determined by treating the substance with 
 a solution of sodium arsenite, acidifying with nitric acid, 
 digesting with an excess of silver nitrate, and finally weighing 
 silver iodide. Carbon and hydrogen were determined by 
 combustion with lead chromate, where the front part of the 
 tube contained a layer of metallic silver which stopped the 
 passage of any iodine. 
 
92 ON SOME COMPOUNDS CONTAINING 
 
 The variations in the results of the analyses are consider- 
 able, and it is probable that the salt, being always deposited 
 in a concentrated mother-liquor, was never quite pure, but 
 there is no evidence that the variations in composition have 
 been regularly influenced by the variations in the conditions 
 of preparation. The analyses are given in the order in which 
 they were made. The last three probably represent better 
 material than the others. 
 
 
 Lead. 
 
 Potassium. 
 
 Iodine. 
 
 Carbon. 
 
 Hydrogen. 
 
 Oxygen 
 (difference.) 
 
 I 
 
 35.51 
 
 4.01 
 
 37.50 
 
 . . . 
 
 ... 
 
 
 II 
 
 36.24 
 
 4.33 
 
 36.16 
 
 ... 
 
 . . 
 
 ... 
 
 III 
 
 35.83 
 
 4.32 
 
 36.01 
 
 . . . 
 
 
 
 
 
 IV 
 
 35.29 
 
 4.07 
 
 37.78 
 
 . . 
 
 . 
 
 
 
 V 
 
 36.21 
 
 4.59 
 
 . . . 
 
 ... 
 
 ... 
 
 . , . 
 
 VI 
 
 35.43 
 
 4.20 
 
 . . . 
 
 . . 
 
 ... 
 
 . 
 
 VII 
 
 35.65 
 
 4.40 
 
 36.49 
 
 ... 
 
 ... 
 
 ... 
 
 VIII 
 
 35.35 
 
 4.15 
 
 . . . 
 
 ... 
 
 ... 
 
 ... 
 
 IX 
 
 34.80 
 
 4.42 
 
 . . . 
 
 . . . 
 
 ... 
 
 ... 
 
 X 
 
 34.85 
 
 3.93 
 
 37.92 
 
 9.14 
 
 1.39 
 
 12.77 
 
 XI 
 
 34.72 
 
 3.97 
 
 39.26 
 
 9.17 
 
 1.41 
 
 11.47 
 
 XII 
 
 34.33 
 
 3.94 
 
 39.83 
 
 8.77 
 
 1.31 
 
 11.82 
 
 Calculated for 5Pb(CH 8 CO 2 ) 2 .3KI.6I, 
 
 35.87 4.07 39.62 8.31 1.04 11.09 
 Johnson found, 
 
 33.195 4.668 43.37 8.63 1.106 9.031 
 
 It must be admitted that the results do not agree very satis- 
 factorily with the calculated quantities, and that the formula 
 is somewhat uncertain. It seems probable, however, that the 
 compound is a combination of lead acetate with potassium 
 triiodide with the formula 5Pb(CH 8 CO 2 ) 2 .3KI 8 . It is not 
 certain that the extra iodine is combined with the potassium 
 rather than with the lead, but since KI 8 is a well-known com- 
 pound, and since the acetic acid radical is present in the 
 proper proportion to form lead acetate, this view seems to be 
 the most plausible one. 
 
LEAD AND EXTRA IODINE. 
 
 93 
 
 Johnson's analysis differs chiefly from the new ones in its 
 higher iodine and consequently lower oxygen as determined 
 by difference. His oxygen is considerably too low for the 
 amount required to give CH 3 CO 2 with the carbon and hydro- 
 gen, and this was evidently the main cause of his inability to 
 arrive at a rational formula. It seems probable that there was 
 an error in his determination of iodine. 
 
 G-roger's Salt A compound has been described by Max 
 Groger * as corresponding to the remarkable formula, PbO. 
 PbI 2 .I 8 . As he prepared it, it was an amorphous precipitate 
 which had been washed with water, and exposed to the air for 
 a long time in order to allow iodine with which it was mixed 
 to evaporate, and, consequently, there seemed to be room for 
 doubt as to its freedom from decomposition after it had under- 
 gone these operations, even if it could be supposed to have 
 been a pure substance when it was precipitated. 
 
 I have undertaken a reinvestigation of this salt, and have 
 succeeded in preparing it in a beautifully crystalline condition 
 in which there was no doubt about its purity, and have found 
 that Groger really analyzed a pure compound, but that he 
 overlooked some water that it contained. With the addition 
 of one molecule of water his formula becomes correct, but this 
 formula, Pb 2 I 6 O.H 2 O, or, as it may be written, Pb 2 l5(OH) 2 , is 
 no less remarkable than the one which Groger advanced. 
 
 This substance, in a crystallized condition, had been ob- 
 served in this laboratory a short time before Gro'ger's work 
 was known here. At my suggestion, Mr. J. H. Pratt had 
 made some experiments with the dark-colored precipitate 
 produced by mixing strong aqueous solutions of lead acetate 
 and potassium triiodide. Such precipitates were collected 
 upon filters, treated while still moist with boiling alcohol, 
 and the resulting liquid, after filtration, was evaporated over 
 sulphuric acid, with the result that small, brilliant black 
 crystals were sometimes obtained. Several partial analyses of 
 this substance showed that it contained lead and iodine in the 
 ratio 2 : 5, and were as follows : 
 
 * Monatshefte fur Chemie, xiii, 610, 1892. 
 
94 ON SOME COMPOUNDS CONTAINING 
 
 n. in. 
 
 Ratio of Calculated for 
 
 average. Pb 2 I 5 O.H 2 O. 
 
 Lead . . . 37.84 . . . 37.32 2.00 38.23 
 
 Iodine . . 57.66 58.71 58.62 5.07 58.63 
 
 The yield of this product was very small, and it was diffi- 
 cult to obtain it in a pure condition, since it was often mixed 
 with the well-known compound PblOH, and with other sub- 
 stances which were not identified. The presence of water in 
 the salt was established, but the circumstances were such that 
 the investigation was interrupted at a point where the pure 
 material at hand had been exhausted, and no accurate determi- 
 nation of water had been made. 
 
 My thanks are due to Mr. Pratt for his valuable assistance 
 in the investigation of the compound up to this point. When 
 I subsequently obtained Groger's salt in a crystallized condi- 
 tion, it proved to have the same form and composition as the 
 product mentioned above, so that a further study of the latter 
 was deemed unnecessary. 
 
 In order to obtain Groger's compound in a well-crystallized 
 condition, it is necessary to modify his method of preparation 
 by using a small amount of acetic acid. It is also advanta- 
 geous to use boiling water instead of cold water for the pre- 
 cipitation, and to use a somewhat larger volume of this than 
 is recommended by him. I have obtained the best results by 
 the following method : Dissolve 10 g. iodine in 100 c. c. 
 absolute alcohol, then 50 g. crystallized lead acetate in 
 150 c. c. water, 3 c. c. glacial acetic acid, and 300 c. c. absolute 
 alcohol. Mix the two solutions, let stand 14 to 16 hours at 
 the temperature of the room, filter to remove the small pre- 
 cipitate, then dilute with 1500 c. c. of boiling water. Let the 
 whole stand until cold, when the compound sought will have 
 crystallized out mixed with iodine. Pour off the liquid and 
 wash the crystals with cold alcohol in small quantities until 
 the iodine is removed. Dry the product upon filter-paper, and 
 then in the air at ordinary temperature. 
 
 The product consists of very brilliant black crystals, usually 
 0.5 mm. or less in diameter. They form octahedra, appar- 
 
LEAD AND EXTRA IODINE. 
 
 95 
 
 ently of the tetragonal system, with faces that are much 
 curved and otherwise distorted. The powder of the crystals 
 is similar in color to Groger's precipitate, and it agrees with it 
 in being practically stable in the air and scarcely acted upon 
 by cold water or alcohol. 
 
 Two separate crops of apparently perfect purity were 
 analyzed. Lead and iodine were determined by the methods 
 described above under Johnson's compound. Water was col- 
 lected and weighed in a calcium chloride tube, the substance 
 being ignited in a tube behind a layer of granulated sodium 
 carbonate which held back the iodine completely. Free iodine 
 was determined volumetrically by the use of sodium thiosul- 
 phate solution. The results were as follows : 
 
 Lead . . . 
 
 Foi 
 L 
 
 . 3854 
 
 ind. 
 
 n. 
 3822 
 
 Calculated for 
 Pb 2 I 6 (OH),. 
 
 Iodine . . . 
 Water . . . 
 Oxygen (diff.) . 
 
 . 58.41 
 . 1.83 
 . 1.42 
 
 58.62 
 1.82 
 1.34 
 
 58.63 
 1.66 
 1.48 
 
 
 100.00 
 
 100.00 
 
 100.00 
 
 " Free " iodine 
 Loss by heating 
 
 34.78 
 36.43 
 
 36.43 I 8 + H 2 
 
 35.18 
 36.84 
 
 I have also prepared the compound, exactly according to 
 Groger's directions, as a reddish-brown precipitate, and after 
 the product was apparently free from intermixed iodine and 
 air-dry, it was dried for three days, spread out in a very thin 
 layer under a bell-jar well charged with solid potassium 
 hydroxide. This product gave the following results on 
 analysis : 
 
 Found. 
 
 Water , 1.80 
 
 Calculated for 
 Pb 2 I 6 (OH),. 
 
 1.66 
 
 This result indicates that Grb'ger overlooked water in his 
 compound, and that his precipitate is identical with the crys- 
 tallized product. 
 
 I have observed the formation of this salt, under various con- 
 ditions, when alcoholic solutions containing lead acetate and 
 
96 COMPOUNDS CONTAINING LEAD AND IODINE. 
 
 iodine, and in some cases potassium iodide also, were diluted, 
 but the purest crops have been obtained only when the ingre- 
 dients were used nearly in the proportion which Grb'ger recom- 
 mends, and also when the alcoholic mixture has been allowed 
 to stand for the proper period. The compound cannot be 
 recrystallized from water, alcohol, or mixtures of the two 
 liquids, and it seems probable, as Groger suggests, that it is 
 formed by the decomposition of some other compound by 
 water. This view does not conflict with the fact that it was 
 prepared, as described above, by the evaporation of certain 
 alcoholic solutions, because these always contained water which 
 increased in proportion to the alcohol as the evaporation went 
 on. The presence of an acetate seems to be indispensable to 
 its production, for I have made a number of experiments using 
 lead nitrate instead of the acetate with no indication of its 
 formation. It seems probable that a soluble compound closely 
 related to Johnson's salt is formed at first, and that this yields 
 Groger's compound by the action of water. 
 
 I have made unsuccessful attempts to prepare a bromide 
 corresponding to Groger's salt, and my attempts to replace a 
 part of the iodine in it by bromine have also failed. 
 
 Conclusion. The two compounds which have been re-inves- 
 tigated, 5Pb(CH 3 CO 2 ) 2 .3KI.6I and PbI 2 .PbO.3I.H 2 O, show 
 no evident relation to each other nor to the compound 2PbI 2 . 
 8KI.L4HiO, which I have previously described, except that 
 all of them are of complicated composition and they all con- 
 tain extra iodine without showing evidence of the existence of 
 lead tetraiodide. Classen and Zahorski's quinoline salt, pre- 
 viously referred to, seems to furnish the only evidence of the 
 existence of this higher iodide in combination. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 March, 1895. 
 
ON THE VOLUMETRIC DETERMINATION OF 
 TITANIC ACID AND IRON IN ORES.* 
 
 BY H. L. WELLS AND W. L. MITCHELL. 
 
 THE difficulties connected with the gravimetric determination 
 of titanic acid make a reliable volumetric method very desir- 
 able, especially for the analysis of titanic iron ores. We have 
 therefore turned our attention to this subject, and have found 
 that satisfactory results can be obtained by a slight modification 
 of a process which has long been known. 
 
 About thirty years ago F. Pisanif stated that the acid under 
 consideration could be determined by reduction with zinc in 
 hydrochloric acid solution, using a gentle heat, and when the 
 violet color no longer deepened, pouring off the liquid from the 
 remaining zinc and titrating with potassium permanganate. 
 Pisani gave no test analyses, and, since his process has not 
 been generally adopted, it is evident that it has not proved 
 satisfactory in the hands of others. 
 
 A number of years ago one of us (Wells) had occasion to 
 analyze a large number of titanic iron ores, and attempted to 
 use Pisani's method with the use of sulphuric acid instead of 
 hydrochloric acid, as recommended by the originator of the 
 process. This modification was made on account of the well- 
 known interference of chlorides with the permanganate method, 
 and it was found that the difficulty mentioned by Pisani, that 
 titanic acid was liable to be precipitated by heating sulphate 
 solutions, could be readily overcome by using a sufficiently 
 large quantity of sulphuric acid. The results of a great many 
 trials at that time, however, showed that the method gave very 
 low results, and the process was then abandoned. The process 
 used in the experiments just referred to was precisely the same 
 
 * Jour. Amer. Chem. Soc., xvii, November, 1895. 
 t Compt. rend., lix, 289. 
 7 
 
98 DETERMINATION OF TITANIC ACID 
 
 as that which we now recommend and which will be described 
 in detail below, except that, after reduction with zinc, the solu- 
 tion was poured off from the excess of that metal into a beaker 
 for titration, an operation which Pisani recommended, and 
 which is customary in the determination of iron by this 
 method. It is now evident that the failure of the method was 
 due to the contact of the solutions with atmospheric air, for, 
 while ferrous sulphate is acted upon very slowly, the sulphate 
 corresponding to the lower oxide of titanium is very rapidly 
 oxidized under such circumstances. 
 
 Marignac,* with his accustomed skill, applied Pisani's 
 method, soon after its publication, to the determination of 
 titanic acid in the presence of niobic acid. He was obliged to 
 use special conditions in order to avoid the reduction of the 
 other acid at the same time, but the feature of his process which 
 is interesting in the present connection is, that he reduced the 
 titanic acid by means of a long rod of pure zinc extending up 
 into the neck of the flask which held the solution, and, after 
 allowing the reduction to take place out of contact with air, 
 he finally took out the zinc and titrated directly in the flask 
 without transferring. Marignac gave a number of test analy- 
 ses which showed that the method gave very good results, 
 although they were a little too low with the larger quantities 
 of titanic acid used. 
 
 We have modified the method of Pisani, as improved by 
 Marignac, by using sulphuric acid solutions and by protecting 
 the liquid during cooling and titration by means of carbon 
 dioxide, and we have also arranged the process for the deter- 
 mination of iron along with the titanic acid. The details of 
 the operation are as follows : 
 
 Five grams of very finely pulverized ore are placed in a rather 
 large beaker, covered with a watch-glass, and treated with about 
 100 c. c. of concentrated hydrochloric acid. A very gentle, 
 gradually increasing heat is applied for several hours, more 
 hydrochloric acid is added if necessary, and, when no further 
 action is apparent, about 50 c. c. of a mixture of equal volumes 
 
 * Zeitschr. anal. Chem., vii, 112. 
 
AND IRON IN ORES. 99 
 
 of concentrated sulphuric acid and water are added, and the 
 whole is evaporated until the sulphuric acid fumes strongly. 
 After cooling, about 200 c. c. of water are added, the whole is 
 heated until the sulphates are dissolved, and the liquid is filtered 
 into a liter flask. With many titanic ores this operation will 
 have dissolved everything except siliceous matter. If, however, 
 some undissolved ore remains, it is ignited, to burn the filter- 
 paper, in a platinum crucible, and the residue is fused with 
 potassium disulphate, at a gradually increasing heat, up to low 
 redness, until the black particles have disappeared. To the 
 cake in the crucible several volumes of concentrated sulphuric 
 acid are added, heat is gradually applied until the whole 
 becomes liquid, then this is heated with a moderate volume of 
 water to dissolve the sulphates, and the liquid is added to the 
 main solution in the liter flask. Filtration may be omitted 
 here, or in the case of the original solution, provided that the 
 siliceous matter is not to be weighed. 
 
 The liquid in the liter flask is diluted to the mark and mixed, 
 and four portions of 200 c. c. each, representing 1 g. of ore, 
 are taken, two of them into Erlenmeyer (conical) flasks of 
 500 c. c. capacity, and the other two into ordinary flasks of 
 350 c. c. capacity. 
 
 To determine iron, hydrogen sulphide is passed into the solu- 
 tions in the ordinary flasks until they are saturated with the 
 gas, then inverted porcelain crucible covers are placed upon 
 the mouths of the flasks, and the solutions are heated and 
 boiled continuously, so that air cannot enter, until the hydro- 
 gen sulphide has been completely removed. This point can 
 be determined by testing the escaping steam with paper which 
 has been dipped in a solution of lead acetate made strongly 
 alkaline with potassium hydroxide. The flasks are then 
 quickly filled to the neck with cold distilled water (which has 
 been recently boiled), best by means of an inverted wash-bottle, 
 directing the stream against the neck of the flask in such a 
 way that the water does not mix to a great extent with the 
 heavier sulphuric acid solution. If the stream of cold water 
 does not strike the top of the neck, there is little danger of 
 
100 DETERMINATION OF TITANIC ACID 
 
 breaking the hot glass. The contents of the flasks are now 
 rapidly cooled by means of a stream of water, transferred to 
 large beakers, and titrated with potassium permanganate 
 solution. 
 
 To the solutions in the Erlenmeyer flasks, about 25 c. c. 
 of concentrated sulphuric acid are added, then, in each case, 
 three or four rods of chemically pure zinc, about 50 mm. long 
 and 6 or 7 mm. in diameter, are attached to the loop of a 
 porcelain crucible cover, which is larger than the mouth of the 
 flask, by means of platinum wire wound securely around them 
 near the middle. The length of the wire is so arranged that 
 the pieces of zinc will be suspended in the liquid when the 
 cover is placed on the flask. When this has been accom- 
 plished, the liquid is boiled gently, so as to keep out air, for 
 thirty or forty minutes, then, without interrupting the boiling, 
 a glass tube, so bent that it extends 50 mm. or more into the 
 flask, and which is deli vering a rather rapid stream of carbon 
 dioxide, is introduced under the cover. Care should be taken 
 to have the carbon dioxide free from air, and that hydrochloric 
 acid which contains sulphur dioxide is not used for its gener- 
 ation. The flask is now rapidly cooled, and then the zinc is 
 washed with a jet of water and removed, and the solution is 
 titrated with permanganate in the flask while the carbon di- 
 oxide is still being passed in. The difference between the 
 permanganate used in this case and that used for the iron 
 alone, represents the amount corresponding to the titanic acid. 
 The factor for metallic iron divided by 0.7 gives the factor for 
 titanic acid (TiO 2 ). 
 
 When a 50 c. c. burette is used, the most convenient 
 strength for the permanganate solution is when 1 c. c. is equal 
 to about 0.014 g. of metallic iron, corresponding to 7.9 g. 
 of potassium permanganate per liter. 
 
 It is customary in this laboratory to standardize permanga- 
 nate solutions by a method which very closely approaches the 
 one described above for the actual determination of iron, so 
 that, if any slight errors are inherent in the process, they are 
 likely to be eliminated because they have an equal effect upon 
 
AND IRON IN ORES. 101 
 
 the standardization and the determination. The method is 
 simple and convenient, and a large amount of experience has 
 shown it to be very accurate. To carry out this operation, a 
 350 c. c. flask is half filled with sulphuric acid (the strong acid 
 diluted with about eight volumes of water). This is heated to 
 boiling with an inverted crucible cover upon the mouth of the 
 flask, and, after the air had been expelled, about 0.6 g. of 
 the purest iron wire, representing nearly the average amount 
 of iron in 1 g. of an ore, is dropped in, and gentle boiling 
 is continued until it has dissolved. The flask is filled to the 
 neck with water, cooled, and finally the liquid is transferred 
 to a beaker and titrated. 
 
 The method of determining iron by reduction with hydrogen 
 sulphide, although well known, does not appear to be as gener- 
 ally used as it deserves to be. The precipitated sulphur present 
 in the liquid has absolutely no effect upon cold permanganate 
 solution, but precipitated sulphides, such as copper sulphide, 
 should be filtered off before boiling. Since concentrated sul- 
 phuric acid is an oxidizing agent, care must be taken to use 
 sufficiently dilute solutions, and not boil them down until the 
 acid becomes strong. 
 
 Potassium titanofluoride Titanium Titanium TJWm. 
 
 taken. found. calculated. 
 
 0.7638 0.1437 0.1527 -0.0090 
 
 0.6425 0.1225 0.1285 -0.0060 
 
 0.7778 0.1524 0.1555 0.0031 
 
 0.6793 0.1308 0.1358 -0.0050 
 
 0.8226 0.1607 0.1645 -0.0038 
 
 1.0956 0.2107 0.2191 -0.0084 
 
 0.4451 0.0848 0.0890 -0.0042 
 
 0.6359 0.1215 0.1271 -0.0056 
 
 0.9004 0.1715 0.1800 -0.0085 
 
 0.4634 0.0882 0.0926 -0.0044 
 
 We have made some test analyses upon the method of de- 
 termining titanic acid volumetrically. Crude potassium titano- 
 fluoride, K 2 TiF, was recrystallized twice from water and used 
 as the source of titanium. Weighed quantities of the care- 
 
102 TITANIC ACID AND IRON IN ORES. 
 
 fully dried salt were evaporated with sulphuric acid, and the 
 resulting substance was treated essentially as has been de- 
 scribed above, but with some variations in the time of boiling, 
 the strength of the acid, and the amount of zinc used. The 
 table on page 101 gives the results obtained in grams. 
 
 The results show a fair degree of uniformity, but they are 
 invariably too low. A part of the deficiency was probably 
 due to the impurities in the potassium titanofluoride used, for 
 it is quite possible that certain impurities may have been in- 
 creased rather than diminished by recrystallizing it, and it is 
 exceedingly difficult to obtain any titanium compound that 
 is certainly free from all other acid-forming elements. The 
 greater portion of the error was doubtless due to the action of 
 air which gained access to the liquid in spite of the precautions 
 used, and it is evident that the accuracy of determinations 
 made by this method would be increased by adding one- 
 twentieth or one-thirtieth to the amount of titanic acid 
 found under the conditions that we have used. 
 
 The great influence of the action of air is shown by two 
 determinations which were made exactly like those given in 
 the preceding table, except that, after cooling in carbon 
 dioxide, the solutions were transferred to beakers and titrated 
 as quickly as possible. 
 
 Potassium titanofluoride Titanium Titanium ,, 
 
 taken. found. calculated. 
 
 0.6831 0.1078 0.1366 0.0288 
 
 0.9545 0.1535 0.1909 0.0374 
 
 The volumetric method, even without correction, will be 
 likely to give more reliable results than those obtained by 
 gravimetric determination, unless great care and skill are 
 displayed in carrying out the latter. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 October, 1895. 
 
ON SOME COMPOUNDS OF TRIVALENT 
 VANADIUM.* 
 
 BY JAMES LOCKE AND GASTON H. EDWARDS. 
 
 THE green solution obtained when vanadic acid is reduced by 
 nascent hydrogen has been but very slightly studied. That 
 it contains salts of vanadium in the trivalent state was recog- 
 nized by Roscoef in the course of his elaborate investigation 
 on the chemical nature of this element. Roscoe, however, 
 made no attempt to study the products which could be obtained 
 from the solution which he prepared by dissolving the anhy- 
 drous chloride, VC1 3 , in water, and failed to make any compari- 
 son between that body and the chlorides of other trivalent 
 elements. 
 
 The first compounds to be isolated from a vanadic solution 
 were prepared by Petersen,J who examined in a very thorough 
 manner the fluoride and its double salts with the fluorides of 
 other metals. His results pointed to a close resemblance 
 between vanadium sesquioxide and its derivatives and the 
 compounds of the groups formed by aluminium, chromium 
 manganese, and iron. Thus, the compound K 2 VF 5 .H 2 O 
 exhibits in its general properties, solubility, etc., close simi- 
 larity to the analogously constituted salts of aluminium, iron, 
 chromium, and manganese. Ammonium vanadifluoride, 
 (NH 4 ) 8 VF 6 , is isomorphous with the ferric salt, (NH 4 ) 8 FeF 6 , 
 described by Marignac, and Petersen prepared other members 
 of the series in (NH 4 ) 8 CrF 6 and (NH 4 ) 8 A1F 6 . A similar 
 relation was observed between double salts with the fluorides 
 of divalent metals, such as CoVF 5 .7H 2 O, CoCrF 6 .7H 2 O, etc. 
 
 * Amer. Chem. Jour., xx, July, 1898. 
 
 t Ann. Chem. (Liebig), Suppl., vii, 78. 
 
 t J. prakt. Chem. (2), xl, 44 (1889). Ann. china, phys., (3) Ix, 306. 
 
104 ON SOME COMPOUNDS OF 
 
 Petersen's work and the conclusions drawn from his results 
 were further substantiated by the recent investigations of 
 Piccini* on the alums of vanadium. He succeeded in isolating 
 the salts NaV(SO 4 ) 2 .12H 2 O, KV(SO 4 ) 2 .12H 2 O, NH 4 V(SO 4 ) 2 . 
 12H 2 0, RbV(S0 4 ) 2 .12H 2 0, CsV(SO 4 ) 2 .12H 2 O, and T1VSO 42 . 
 12H 2 O. With the exception of a short article by Brieiiy,f 
 who prepared a vanadium sulphuric acid, or ' alum acid,' J 
 V(SO 4 )SO 4 H.4H2O, the above investigations embrace practi- 
 cally all that has been published on vanadic salts. 
 
 We have recently undertaken the preparation of a number 
 of other compounds, the analogues of which are of character- 
 istic nature in the case of chromium, in the hope of ascertain- 
 ing more definitely the influence which the atomic weight of 
 vanadium exerts upon the development of the properties com- 
 mon to the compounds of the group : aluminium, vanadium, 
 chromium, manganese, iron, and cobalt. 
 
 The chief difficulty which an investigation of this kind pre- 
 sents lies in the extreme readiness with which vanadic 
 solutions absorb oxygen, with formation of vanadyl salts. 
 Petersen was able to start directly from the anhydrous sesqui- 
 oxide, which is soluble in hydrofluoric acid. The solutions of 
 vanadic sulphate used by Piccini in the preparation of the 
 alums were obtained simply by the electrolysis of vanadic acid 
 in a solution of sulphuric acid. These methods, while sat- 
 isfactory in individual cases, are of course limited in their 
 applicability, and they could not be used in the preparation 
 of such compounds as a vanadicyanide, sulphocyanate, or the 
 like. We were therefore compelled to start out from the 
 readily oxidizable vanadic hydroxide, precipitated by an alkali 
 after the reduction of the pentoxide with sodium amalgam. 
 
 In order to protect the hydroxide and solutions from oxi- 
 dation, all operations were carried out in an atmosphere 
 of hydrogen. For this purpose the apparatus shown in the 
 
 * Zeitschr, anorg. Chem., xi, 106 ; xiii, 441. 
 t J. Chem. Soc. (London), xlix, 822. 
 
 t Chromium is the only other alum-forming metal which yields such an 
 acid, Cr(S0 4 )S0 4 H. 
 
TRIVALENT VANADIUM. 
 
 105 
 
 accompanying figure was employed. The pear-shaped bulb 
 a, of which we had several pieces, holds about 500 c. c. Over 
 
 its drawn-out end, which is two-thirds of an inch in diameter, 
 passes a piece of thick-walled, soft-rubber tubing, which can 
 be closed by a stop-cock e. The tube b is of capillary diameter, 
 fitted with a glass cock, and bent over on itself, to more 
 securely prevent the entrance of air when this cock is open, 
 c is a somewhat wider tube, which serves for the introduction 
 of reagents, and is closed by a stop-cock when necessary. 
 
 For heating on the water-bath, the bulb is placed on the 
 latter, mouth downward, while a rapid current of hydrogen is 
 passed in at <?, a being open. The contents of the bulb may 
 be boiled in a similar manner, the bulb then resting on its 
 
106 ON SOME COMPOUNDS OF 
 
 side. Reagents are added by means of a small flask m, fitted 
 with a cork holding two short tubes, one of which is connected 
 with the hydrogen generator. The reagent having been intro- 
 duced into the flask (a test-tube is also convenient) the air in 
 the latter is displaced by hydrogen, and the second tube then 
 connected with <?. The flask is then simply inverted, where- 
 upon the reagent runs into the bulb. 
 
 The most important and difficult operation involved in the 
 work was of course the filtration and washing of the vanadic 
 hydroxide. To perform this without exposing the substance 
 to the air, a Buchner's funnel was used, to which was joined by 
 a wide rubber band a well-fitting piece of apparatus, d, like an 
 inverted funnel. The stem of the latter was of the same 
 diameter as the mouth of the bulb. Before attaching this 
 filtering-apparatus to the bulb, it was entirely filled, together 
 with the suction-flask, with water, and the latter then dis- 
 placed by hydrogen. Connection was then made, slight 
 suction applied, and the stop-cock at the mouth of the bulb 
 opened. The precipitate was washed with water from which 
 the air had been expelled by boiling, from a flask connected as 
 for the introduction of reagents. 
 
 When the next operation involved the solution of the pre- 
 cipitate in an acid, the filtrate was drawn off by suction 
 through the pump, the suction-flask rinsed with the water from 
 one or two additional washings, and the acid then introduced 
 as above. The solution was then transferred, either to another 
 bulb, or, if it was to be evaporated to crystallization, to a 
 crystallizing-dish, the side tube of the suction-flask being in 
 that case held below the surface of some benzine placed in 
 the dish. 
 
 By taking proper care in the observance of minor details, 
 such as the filling of the tubes with water before making connec- 
 tions, etc., we were able almost entirely to obviate the danger 
 of oxidation ; and after a little practice we could carry out the 
 operations of filtration and the like almost as rapidly as in 
 the open air. In one afternoon, starting out with vanadium 
 pentoxide, we have prepared vanadium dihydroxide, V(OH) 2 , 
 
TRIVALENT VANADIUM. 
 
 107 
 
 washed it, redissolved in hydrochloric acid, and brought the 
 solution into a desiccator. In spite of all the operations sub- 
 sequent to the reduction of the vanadic acid, the final solution 
 possessed the true lavender color of vanadious salts, without 
 a trace of green or brown. Roscoe describes such a solution 
 as being a more sensitive reagent towards oxygen than is 
 pyrogallol itself. 
 
 The vanadium preparations which we used in this work 
 were placed at our disposal through the kindness of Prof. 
 Paul Jannasch, of Heidelberg, Germany. They consisted 
 chiefly of thoroughly purified vanadates of sodium and am- 
 monium. These compounds were worked up as follows : 
 Enough of the substance to yield about 5 g. of vanadium 
 hydroxide was dissolved in a small quantity of water, 10 c. c. 
 of concentrated hydrochloric acid were added, and the solu- 
 tion boiled with alcohol to reduce the vanadic acid to vanadyl 
 dichloride, VOC1 2 . After the alcohol had been driven off, 
 the solution was transferred to a bulb, and while a rapid cur- 
 rent of hydrogen was led through the latter, 5 per cent 
 sodium amalgam was gradually introduced hi small lumps, 
 the solution being in the meantime kept acid by the occa- 
 sional addition of hydrochloric acid. The reduction was con- 
 tinued until the solution just began to lose the pure green 
 color of the vanadic salts and assume a bluish-green tint, due 
 to compounds of the next lower degree of oxidation. The 
 operation required the addition of about 700 g. of amalgam. 
 
 The mercury was next drawn off, the solution filtered, 
 transferred to another bulb, and treated in the cold with just 
 enough ammonium hydroxide to precipitate the vanadic hy- 
 droxide * completely. The latter comes down as a dirty green 
 flocculent precipitate, which absorbs oxygen with the greatest 
 avidity. It was allowed to stand for some tune, and then fil- 
 tered and washed thoroughly with warm water from which 
 the air had been expelled. From this precipitate the follow- 
 ing salts were obtained by solution and crystallization. 
 
 * Potassium hydroxide does not work as well for this purpose, as it dis- 
 solves more or less of the vanadic hydroxide. 
 
108 ON SOME COMPOUNDS OF 
 
 Vanadium Trichloride, VCl^GH^. 0. Halberstadt * men- 
 tions the fact that when the anhydrous chloride, VC1 3 , is dis- 
 solved in water and the solution evaporated over sulphuric 
 acid, a very unstable crystalline compound is obtained. Pic- 
 cini mentions in a foot-note to his first article on the vana- 
 dium alums f that he had obtained this substance in distinct 
 crystals, and found it to have the above composition. This 
 foot-note escaped our notice when first reading his article, and 
 at the time of preparing the compound we supposed we were 
 the first to have isolated it. As, however, nearly three years 
 have elapsed since his article was sent in for publication, we 
 may be allowed to describe the compound, yielding to him the 
 priority of its discovery. 
 
 It is obtained by dissolving vanadic hydroxide in concen- 
 trated hydrochloric acid and evaporating the green solution 
 to dryness in a vacuum-desiccator. The salt separates out 
 from the syrupy liquid in large green prisms, some of which 
 attained with us the length of nearly half a centimeter. It 
 dissolves in water with extreme readiness, yielding, like the 
 other neutral vanadic salts, a brown solution which becomes 
 green on acidification. The salt is very deliquescent, and on 
 exposure to the air for any length of time, dissolves in the 
 water absorbed and is oxidized to vanadyl dichloride. It is 
 readily soluble in both alcohol and ether, but no distinct crys- 
 tals could be obtained from its solution in these liquids. 
 
 In the analysis of the substance the solution was acidified 
 with nitric acid and the chlorine precipitated with silver nitrate. 
 The vanadium was determined in another portion by titra- 
 tion from the tetravalent to the pentavalent state with iodine, 
 according to Browning's J method. The water was estimated 
 by difference : - 
 
 Calculated for ., , 
 
 VC1 3 .6H 2 0. Found. 
 
 V 19.24 18.96 
 
 Cl .... 40.10 39.95 
 
 H 2 .... 40.66 41.09 
 
 100.00 100.00 
 
 * Ber. d. chem. Ges., xv, 1619 (1882). 
 
 t Zeitschr. anorg. Chem., xi, 107 (1896). | Ibid., i, 158. 
 
TRIVALENT VANADIUM. 
 
 109 
 
 An attempt was made to measure the crystals, but they 
 proved too hygroscopic. Their optical properties, however, 
 were found to conform to the rhombic system. Salts of com- 
 position similar to that of this compound are seen in A1C1 3 . 
 6H 2 O, CrCl 3 .6H 2 O, and FeCl 3 .6H 2 O. 
 
 Potassium Vanadichloride. An attempt was made to pre- 
 pare from the chloride a double salt analogous to those of the 
 series R 2 FeCl 5 .H 2 O, in which R is K, Rb, NH 4 , etc. A few 
 grams of the trichloride were dissolved in concentrated hydro- 
 chloric acid, the calculated quantity of potassium chloride added, 
 and the mixture left to crystallize in a vacuum. The product 
 consisted chiefly of green crystals of a somewhat lighter shade 
 than that of the pure vanadium chloride, but it was impos- 
 sible to isolate these completely. A vanadium determination, 
 made in as pure a product as we could obtain, showed that it 
 contained 22.41 per cent V. The quantity calculated for the 
 anhydrous compound KVC1 4 is 22.03 per cent V. 
 
 Vanadium Bromide, VBr 3 .6H 2 0. This compound was 
 prepared in a manner strictly analogous to that by which the 
 chloride was obtained, pure concentrated hydrobromic acid 
 being used. It crystallizes with less readiness than the 
 chloride, and decomposes more easily. In other respects the 
 two compounds were closely similar. The bromide decom- 
 poses more or less on solution in water, leaving as a residue 
 a small quantity of a brown substance, probably a basic bro- 
 mide. Like the chloride, it is soluble in both alcohol and 
 ether, to a green solution. The analysis was carried out as 
 in the case of the chloride. 
 
 V . 
 Br 
 
 H 2 
 
 Calculated for 
 VBr 8 -f- 6H 2 O. 
 
 , 12.83 
 60.10 
 27.07 
 
 100.00 
 
 The iodide could not be obtained. Vanadium hydroxide 
 dissolves in hydriodic acid as readily as in hydrochloric or 
 hydrobromic, but the solution turns brown on evaporation 
 
110 ON SOME COMPOUNDS OF 
 
 and eventually leaves only an amorphous, brownish-black 
 residue, only partially soluble in water. 
 
 Potassium Vanadicyanide, K z V( CN^) 6 . For the preparation 
 of this compound, it was found more convenient to start out 
 from the anhydrous vanadium trichloride, which we prepared 
 according to the method of Halberstadt.* About 5 g. of 
 this substance were dissolved in as little water as possible f 
 and slightly acidified with hydrochloric acid. A concentrated 
 solution of potassium cyanide containing about one and a half 
 times the calculated quantity of the salt was placed hi a bulb, 
 and the vanadium chloride solution then added. The mixture 
 at once assumed the form of a thick, deep-purple paste which 
 gradually became thin again, though without at first losing its 
 color, and remaining almost opaque. This part of the reaction 
 was observed by Petersen,J who states that he thus obtained 
 a dark blue solution. The blue color, however, is in fact due 
 only to very finely divided particles of the original precipitate 
 suspended in the solution, which is itself of a deep wine color. 
 After shaking for some time, the liquid cleared, and only a 
 few flakes of a brown residue, presumably the hydroxide, 
 remained undissolved. It is absolutely necessary to have a 
 considerable excess of potassium cyanide present, as the pre- 
 cipitate dissolves with great difficulty, and a clear solution 
 cannot otherwise be obtained. 
 
 The wine-colored solution, after being filtered, was treated 
 with just enough alcohol to bring about incipient precipitation, 
 and then allowed to stand for some hours surrounded by ice- 
 water. A fine precipitate separated out, which consisted of a 
 mixture of potassium cyanide and vanadicyanide, and in addi- 
 tion to this, comparatively large crystals of the latter collected 
 on the sides and bottom of the vessel. These alone were 
 
 * Ber. d. chem. Ges., xv, 1619 (1882). 
 
 t The formation of the hydrated chloride is readily seen when a quantity 
 of the anhydrous chloride is added to about its own volume of water. It dis- 
 solves with a hissing sound, and, on cooling, the liquid solidifies to a green 
 crystalline mass. On addition of more water, this dissolves to a brown 
 solution. 
 
 J J. prakt. Chem., xl, 50. 
 
TRIVALENT VANADIUM. 
 
 Ill 
 
 saved, the rest of the product being removed by lixiviation. 
 The crystals were repeatedly washed by decantation with 95 
 per cent alcohol, and finally with ether, and dried in a 
 vacuum-desiccator. 
 
 In the analysis of the product the carbon and nitrogen were 
 determined by combustion. The vanadium was estimated by 
 titration with iodine, and the potassium as sulphate, in a sepa- 
 rate portion, after the oxidation of the vanadium to vanadic 
 acid and its removal as lead vanadate.* 
 
 V 
 
 c 
 
 N 
 K 
 
 Calculated for 
 
 K 3 V(CN) 6 . 
 
 . 15.74 
 . 22.22 
 . 25.93 
 . 36.11 
 100.00 
 
 Pound. 
 
 15.89 
 21.80 
 26.36 
 36.47 
 100.52 
 
 This salt forms another member of the series of complex 
 cyanides of the formula K 3 M(CN) 6 , of which the other mem- 
 bers as yet known are K 8 Cr(CN) 6 , K 8 Mn(CN) 6 , K 8 Fe(CN) 6 , 
 K 3 Co(CN) 6 , K 8 Rh(CN) 6 , and K 3 Ir(CN) 6 . The crystals ob- 
 tained were about a millimeter in length, and of a bright 
 scarlet color. Owing to their instability, it was impossible 
 to measure them, and thus determine whether they were 
 isomorphous with the other members of the series. They 
 appeared under the microscope to be rhombic plates, with 
 well-formed domes and base ; in polarized light, however, they 
 showed inclined extinction, and are therefore probably mono- 
 clinic like the others. In potassium ferricyanide the angle 
 is 90 6'. 
 
 Potassium vanadicyanide is readily soluble in water, in- 
 soluble in alcohol. Its aqueous solution grows turbid within 
 a few minutes, however, but is much more stable when con- 
 taining some free potassium cyanide. Even in that case it 
 cannot be kept for any length of time. The freshly-prepared 
 solution is at once decomposed by acids, turning green. It 
 
 * Roscoe : Ann. Chera. (Liebig), Suppl., viii, 102. 
 
112 ON SOME COMPOUNDS OF 
 
 gives off a slight odor of hydrocyanic acid, as does the solid 
 salt itself. The solution is at first stable toward alkalies in 
 the cold, but on heating or standing for some time the hy- 
 droxide separates out. 
 
 The solution yields colored precipitates with the neutral 
 solutions of various metals, of which the following are the 
 most distinct : 
 
 Ferrous iron .... red-brown. 
 
 Cadmium yellow. 
 
 Copper yellow. 
 
 Nickel purple. 
 
 Manganese .... greenish-yellow. 
 
 Silver and mercury salts are reduced by it, with deposition 
 of the metals. None of these precipitates is stable toward 
 acids, and their color soon undergoes a change on standing. 
 
 We have made repeated attempts to isolate the purple pre- 
 cipitate which separates on the first addition of potassium 
 cyanide to the vanadic solution, but without success. The 
 compound, probably vanadic cyanide, is extremely unstable, 
 and on drying yields a green or brown amorphous product, 
 which is obviously a mixture. 
 
 The vanadicyanides of ammonium and sodium seem to exist 
 only in solution. Vanadium cyanide dissolves in excess of 
 ammonium cyanide or sodium cyanide, to solutions of the 
 same color as that of the potassium salt. No crystalline 
 products, however, could be obtained from either solution, 
 either by evaporation or precipitation with alcohol. When 
 the latter is employed, the compounds decompose at once, 
 with separation of a thick blue paste. 
 
 The properties and reactions of potassium vanadicyanide 
 are of special interest in view of the relative stability of the 
 complex cyanides of the other metals of the group. The only 
 stable compounds of trivalent cobalt are those which contain 
 the metal as a constituent of a complex radical, either posi- 
 tive, as in the cobaltiamines, or negative, as in K 3 Co(CN) 6 , 
 H 3 Co(CN) 6 , etc. The corresponding ferric compounds, in 
 
TRIVALENT VANADIUM. 
 
 113 
 
 comparison with other ferric salts, are somewhat less stable 
 than the cobaltic compounds as compared with simple cobaltic 
 salts. Thus, for example, potassium ferricyanide is less 
 stable, compared with ferric sulphate, than is potassium 
 cobalticyanide, when compared with cobaltic sulphate. The 
 complex manganese derivatives are relatively still less stable. 
 Among the latter is a sodium salt, Na 8 Mn(CN) 6 , but the free 
 acid is unknown. Chromic cyanide yields neither an acid, 
 H 3 Cr(CN) 6 , nor a sodium salt, and ammonium chromicyanide * 
 is very unstable. The chromicyanide solutions, however, 
 are stable towards alkalies even on boiling. In the case of 
 vanadium, neither the sodium nor ammonium salts can be 
 obtained; potassium vanadicyanide is instantly decomposed 
 by acids, with evolution of hydrocyanic acid, and is stable 
 towards alkalies only in the cold. The simple vanadic salts 
 are comparatively stable. Aluminium, which has the lowest 
 atomic weight of all the metals in the group, is precipitated 
 as hydroxide when potassium cyanide is added to its solution, 
 and no cyanogen compounds at all of this metal can be ob- 
 tained. The tendency to form complex radicals, throughout 
 the entire group, as compared with the tendency to form 
 simple salts, is thus seen steadily to diminish with a decrease 
 in the atomic weights of its members. 
 
 Potassium Vanadisulphoeyanate, K^V^ONS)^H t O. 
 Among the characteristic compounds of trivalent chromium 
 the derivatives of chromic sulphocyanate are very prominent. 
 Potassium chromisulphocyanate, K 3 Cr(CNS).6H 2 O, is almost 
 as stable as the chromicyanide. It is not decomposed by 
 either alkalies or acids in cold solution, f A number of other 
 salts, such as Ag 3 Cr(CNS) 6 , Ba3[Cr(CNS) 6 ] 2 , etc., derivatives 
 of the same acid, H 3 Cr(CNS) 6 , are also known. 
 
 We have succeeded in preparing a compound of vanadium 
 analogous to this potassium salt, and find that it corresponds 
 very closely to the latter in its reactions. The method em- 
 ployed was as follows : 
 
 An alcoholic solution of potassium sulphocyanate was 
 
 * Ann. Chera. (Liebig), iii, 163. 
 
 t Rosier: Ibid., cxli, 185. 
 
 8 
 
114 ON SOME COMPOUNDS OF 
 
 made by fusing sulphur with potassium cyanide,* digesting 
 the product with absolute alcohol, and filtering. To this 
 solution was added somewhat less than the calculated quan- 
 tity of vanadium chloride, dissolved in a small volume of 
 water. A precipitate of potassium chloride at once appeared, 
 and the solution assumed a deep brown color. After digestion 
 for an hour on the water-bath, the solution was concen- 
 trated by evaporation, and then placed in a vacuum-desicca- 
 tor to crystallize. The first crop of crystals consisted of a 
 mixture of about equal proportions of potassium sulphocy- 
 anate and vanadisulphocyanate. This was removed, and the 
 evaporation continued until the solution was of a thick, syrupy 
 consistency. A large quantity of homogeneous, dark-red crys- 
 tals, almost black, were thus obtained. They were cleaned 
 as thoroughly as possible by pressure between filter paper, 
 washed with ether, and dried in a vacuum. The analysis 
 gave the following results : 
 
 Calculated for w/ . ^ 
 
 K 3 V(CNS) 6 .4H 2 0. 
 
 K 19.90 19.52 
 
 C 12.24 
 
 N 14.29 14.73 
 
 S 32.65 33.23 
 
 V 8.68 8.55, 8.22, 8.79 
 
 H 2 . . . . 12.24 13.09 
 100.00 
 
 Potassium vanadisulphocyanate, like the corresponding 
 chromium salt, is extremely soluble in both alcohol and water, 
 but is stable only in presence of an excess of potassium sul- 
 phocyanate. The pure salt is decomposed by either solvent, 
 forming a green solution. Crystals mixed with a small quan- 
 tity of sulphocyanate are very hygroscopic, dissolving in the 
 water absorbed. Toward oxygen the salt is the most stable 
 of any which we, have prepared. The vanadium in the radi- 
 cal V(CNS) 6 undergoes oxidation only very slowly, and in 
 
 * Chem. Zeitung, 1866, 666. 
 
TRIVALENT VANADIUM. 
 
 115 
 
 presence of potassium sulphocyanate the solution may be left 
 exposed to the air for some time without losing its charac- 
 teristic dark-brown color. Alkalies precipitate vanadic hy- 
 droxide from the solution only on boiling, but it is at once 
 decomposed by acids. 
 
 The preparation of corresponding salts of other metals, such 
 as Na 3 V(CNS) 6 , Ba 8 [V(CNS) 6 ] 2 , etc., we have not yet 
 attempted, but we hope to do so in the near future. Our in- 
 vestigations on the vanadic compounds in general will be 
 continued. 
 
 NEW HAVEN, May, 1898. 
 
ON AN ISOMER OF POTASSIUM FERRICYANIDE.* 
 
 BY JAMES LOCKE AND GASTON H. EDWARDS. 
 
 THE behavior of potassium ferricyanide toward powerful 
 oxidizing agents has been virtually but once the subject of 
 investigation. In 1869 Stadeler,f in support of his theory 
 that the iron in sodium nitroprussiate, Na 2 Fe(CN) 5 NO 2 , is 
 tetravalent, sought to obtain the related perferricyanide, 
 K 2 Fe(CN) 6 , by the action of iodine upon the ferricyanide. 
 He thus obtained a greenish-brown, crystalline product, which 
 was apparently too impure for analysis, for he assigned the 
 above formula to it without having quantitative data upon 
 which to base his conclusions. A body having approximately 
 the same characteristics as Stadeler's compound was after- 
 ward obtained by Bong,! by the action of potassium chlorate 
 and sulphuric acid upon the ferricyanide. To this product 
 was likewise assigned the formula K 2 Fe(CN) 6 , but it could be 
 prepared only in admixture with a large percentage of potas- 
 sium sulphate, and no analysis of it was made. The principal 
 work upon the subject, and the only work in which analytical 
 results were obtained, was performed by Skraup in 1877. 
 The latter, to a certain extent, adopted Bong's method, but 
 he substituted hydrochloric acid for sulphuric, and isolated 
 his reaction-product by repeatedly precipitating the aqueous 
 solution with alcohol. He finally obtained a completely 
 amorphous, black or dark-violet powder, which was intensely 
 hygroscopic and smelled strongly of cyanogen. This body 
 Skraup submitted to thorough analysis, but he was unable 
 to obtain satisfactory results. The percentage of cyanogen 
 fell much below the amount calculated for the compound 
 
 * Amer. Chem. Jour., xxi, March, 1899. 
 
 t Ann. Chem. (Liebig), cli, 1. 
 
 J Bull. Soc. Chim. (1875), xxiv, 26& 
 
AN ISOMER OF POTASSIUM FERPJCYANIDE. 117 
 
 K 2 Fe(CN) 6 (about 4.0 per cent). But the iron and potassium, 
 while correspondingly high, were present in the approximate 
 ratio of one to two, and he therefore assumed the body 
 possessed the formula previously suggested by Stadeler and 
 Bong. 
 
 The details of the method of preparation used by Skraup 
 were briefly as follows : Assuming the reaction to proceed in 
 the simplest manner, viz., according to the equation, 
 
 6K 3 Fe(CN) 6 + KC10 3 + 6HC1 = 6K 2 Fe(CN) 6 + 7KC1 + 3H 2 0, 
 
 he added to the hot solution of 50 g. potassium ferricyanide 
 and 4 g. of potassium chlorate, in 100 c. c. water, the cal- 
 culated quantity of hydrochloric acid (4 g., sp. gr. 1.19) in 
 about 75 c. c. water. The solution quickly assumed a pecu- 
 liar red color, and after a few minutes effervescence was 
 observed, presumably escaping cyanogen chloride. Shortly 
 afterwards the solution was cooled down and allowed to 
 stand for twenty-four hours. Precipitation with alcohol then 
 yielded a crystalline product, which was redissolved in water 
 and reprecipitated with alcohol about twelve tunes. On 
 the third or fourth repetition of this operation the body 
 began to lose its crystalline nature, and the final product, 
 which he used for his analyses, was completely amorphous. 
 The most characteristic reaction of the new compound, and 
 the only one which indicated that its iron was in the tetra- 
 valent state, was found to be its decomposition by alkalies. 
 On being boiled with the latter it yielded ferric hydroxide, 
 potassium ferrocyanide, and potassium cyanate. 
 
 The method employed by Skraup is open to criticism at two 
 points. If the reaction proceeds as he supposed, it should be 
 stopped before the evolution of cyanogen chloride begins, for 
 this must be due to the decomposition of the substance. In 
 describing the reactions of the salt, Skraup states that it is 
 slowly decomposed by alcohol. It would hardly seem advisa- 
 ble, therefore, to employ repeated precipitation with alcohol 
 as a means of purifying it. These considerations led us to 
 believe that the final, amorphous body was simply a decompo- 
 
118 ON AN ISOMER OF 
 
 sition-product of the supposedly impure crystalline precipitate 
 obtained on the first addition of alcohol to the oxidized solu- 
 tion, and that this body might perhaps be obtained in a state 
 more suitable for analysis by some other supplementary 
 process. 
 
 We therefore prepared this salt according to Skraup's 
 directions, but placed the mixture in ice-water as soon as the 
 first sign of effervescence was observed. The time required 
 for the reaction varies greatly with the temperature. When 
 the hydrochloric acid is added to the boiling solution, the gas 
 comes off at once, but at 95, the temperature at which we 
 worked, about five minutes are required. When the solution 
 had cooled to almost 20, it was filtered, and slightly less than 
 an equal volume of alcohol, or just enough to bring about 
 incipient precipitation, was added. The solution was then 
 allowed to stand until its temperature had fallen nearly to 0. 
 A dense crystalline precipitate separated out, from which the 
 mother-liquid was drawn off as completely as possible, without 
 washing, on a suction filter. The product was then redissolved 
 in as little water as possible, and partially reprecipitated with 
 a much smaller volume of alcohol. On a second repetition of 
 this operation, the crystalline nature of the precipitate became 
 less distinct, as Skraup also observed. The purification was 
 therefore carried no further. The salt was extremely soluble 
 in water, yielding a solution which was green by reflected 
 light, and by transmitted light had a peculiar reddish tint. 
 Thus far it corresponded with Skraup's observations, but on 
 boiling with ammonium hydroxide it gave only a trace of 
 ferric hydroxide and potassium ferrocyanide,* indicating that 
 our supposition was correct, that Skraup's body was a product 
 of decomposition of his first precipitate. 
 
 The salt obtained on the third precipitation with alcohol 
 (the yield was about 1 g.) was carefully examined under 
 the miscroscope for impurities. It consisted of very small, 
 
 * A quantitative determination of the ferrocyanide thus formed by 
 titration with potassium *permanganate, showed that only 0.07 per cent of 
 the salt had undergone reduction. 
 
POTASSIUM FERRICYANIDE. 
 
 119 
 
 greenish-yellow needles, among which no foreign ingredients 
 could be seen. The body was, therefore, subjected to analy- 
 sis, in the expectation that it would give results correspond- 
 ing closely with the formula K 2 Fe(CN) 6 . To our surprise, 
 however, we found that, together with 11.83 per cent of 
 water, it contained the four elements in the same ratio as 
 potassium ferricyanide itself, corresponding almost exactly 
 with the formula K 3 Fe(CN) 6 . The results of this first analy- 
 sis (I) follow : 
 
 H 2 . . 
 Fe. . . 
 
 Calculated for 
 K 3 Fe(CN) 6 with 
 11.83 per cent H 2 O 
 
 . . 11.83 
 . . 15.01 
 
 Found. 
 
 11.83 
 15.43 
 19.20 
 
 22.70 
 31.60 
 
 Ratio. 
 
 0.344 
 1.97 
 2.07 
 1.0 
 
 1.03 
 5.91 
 6.21 
 3.0 
 
 C . . . 
 
 . . 19.30 
 
 N . . . 
 
 . . 22.51 
 
 K . . . 
 
 . . 31.35 
 
 
 
 100.00 
 
 100.76 
 
 The water present we afterwards found to be due chiefly to 
 the extreme difficulty with which the salt can be dried. On 
 precipitation the latter comes down as a network of very fine, 
 delicate needles, in the capillary spaces between which the 
 water is most obstinately retained. By allowing the salt to 
 stand for several days in a vacuum over sulphuric acid, the 
 percentage of water can be reduced until it corresponds with 
 the value calculated for the formula K 3 Fe(CN) 6 +H 2 O.* Our 
 first supposition was that the presence of this water was the 
 cause of the difference between the new compound and the 
 normal ferricyanide, or, in other words, that it was water of 
 constitution. This hypothesis, however, is inadmissible ; for 
 the corresponding silver salt, which can readily be obtained, is 
 like normal silver ferricyanide, an anhydrous compound, and 
 has the formula Ag 3 Fe(CN) 6 . The salt must, therefore, be 
 regarded as an actual isomer of potassium ferricyanide.* What 
 the structural difference between the two is, or how the isomer 
 happens to be formed through the action of such a substance 
 
 * Found, 5.55 per cent. The theory requires 5.18 per cent. 
 
120 ON AN ISOMER OF 
 
 as chloric acid, we cannot explain. But repeated analyses of 
 the compound itself, and also of the silver salt, together with 
 the quantitative study of reactions in which it was completely 
 converted into the normal ferricyanide, leave no doubt as to 
 its composition. Its reactions, on the other hand, are in some 
 cases totally different from those of the latter compound. We 
 propose for it, as a temporary designation, the term potassium 
 yQ-ferricyanide. 
 
 The various analyses of the isomer published below were 
 made from as many different preparations obtained under as 
 varied conditions as possible. In the course of the work it 
 was found that two precipitations with alcohol are sufficient 
 to yield the body in a state pure enough for analysis. Even 
 on the first precipitation it is nearly pure, giving no ferric 
 hydroxide when boiled with ammonia and containing only a 
 trace (about 0.5 per cent) of chlorine. By using somewhat 
 less than an equal volume of alcohol on the second precipita- 
 tion, a yield of about 15 g. could be obtained, and when 
 the conditions were very carefully observed the individual 
 crystals were large enough to be observed by the naked eye. 
 A slight variation of the conditions in any particular, how- 
 ever, leads to the formation simply of a crystalline paste, 
 which it is almost impossible to purify. The properly pre- 
 pared salt can be easily and thoroughly washed with 75 per 
 cent alcohol; when placed in a vacuum, after subsequent 
 washing with absolute alcohol, it falls to an extremely light, 
 voluminous powder. This possesses a pure, rich olive color, 
 which appears brown when observed by gaslight. It dis- 
 solves with the utmost ease in water, but, unlike Skraup's 
 compound, it is not noticeably hygroscopic. The solution is 
 comparatively stable, though it undergoes gradual decompo- 
 sition on standing. This decomposition is not accompanied 
 by the formation of either potassium cyanide, hydrocyanic 
 acid, or free cyanogen, nor does the dry salt possess the 
 slightest odor of the latter substance. 
 
 Various attempts were made to obtain the salt in the form 
 of larger crystals, by the evaporation of its concentrated 
 
POTASSIUM FERRICYAN1DE. 
 
 121 
 
 aqueous solution. But owing to the capillary action of the 
 needles as they separated on the sides of the vessel, only a 
 dense efflorescent growth was obtained, from which no indi- 
 vidual crystals could be isolated. Analysis II below is from 
 one of these crops. When a small quantity of the salt is dis- 
 solved in a drop or two of water and allowed to crystallize out 
 on an object glass, it is obtained as an intimate network of 
 microscopic needles, of characteristic, very slightly tapering 
 form. They are probably rhombic, showing parallel extinc- 
 tion and slightly pleochroic. Among them no trace of the 
 heavy prismatic or block-shaped crystals of the normal or 
 a-ferricyanide could be detected. In crsytallization experi- 
 ments with mixtures of the two isomers, on the other hand, 
 they could be distinguished at a glance, showing that the new 
 compound is not merely the a-fenicyanide in a new crystallo- 
 graphic modification. 
 
 Analytical Results. For the determination of potassium 
 and iron the compound was decomposed by heating with con- 
 centrated nitric acid. The iron was then precipitated with 
 ammonium hydroxide, and the potassium weighed as chloride 
 or sulphate. The nitrogen was determined by combustion. 
 The total combustion of the carbon takes place with great 
 difficulty, and the results for that element were rather low 
 (one per cent or more). In view of the above determination, 
 (I) however, together with experiments on the quantitative 
 conversion of the compound into potassium a-ferricyanide, its 
 further determination was deemed unnecessary, and the per- 
 centage of cyanogen was calculated from that of the nitrogen. 
 The water was determined by combustion. 
 
 ii. 
 
 in. 
 
 CN 
 
 Fe . 
 K . 
 H 2 
 
 Found. 
 Per cent. 
 
 45.31 
 
 16.95 
 
 33.91 
 
 5.41 
 
 101.47 
 
 Ratio. 
 
 6.00 
 1.04 
 3.00 
 1.00 
 
 Pound. 
 Per cent. 
 
 45.13 
 
 16.02 
 
 32.90 
 
 4.65 
 
 98.70 
 
 Ratio. 
 
 6.15 
 1.01 
 3.00 
 0.8 
 
122 ON AN ISOMER OF 
 
 IV. 
 
 Ratio Average of tho Calculated for 
 
 3 Analyses. K 3 Fe(CN) 6 .H 2 O. 
 
 CN . . 45.05 5.94 
 
 Fe . . 16.98 1.05 
 
 K . . . 34.11 3.00 
 
 H 2 . . 4.73 0.87 
 100.87 
 
 A subsequent determination of water, according to the 
 method of Jannasch and Locke,* in another preparation which 
 had previously been allowed to stand in a vacuum over sul- 
 phuric acid for three days, gave 5.55 per cent. There seems 
 to be no doubt, therefore, that the body possesses the formula 
 K 8 Fe(CN) 6 .H 2 O. In order to prove this definitely, however, 
 and at the same time make a direct estimation of the cy- 
 anogen, a method was sought by which the body could be 
 quantitatively converted to the a-ferricyanide. Preliminary 
 experiments showed that it passed into potassium ferrocyanide 
 upon reduction, but its titration on that principle, either ac- 
 cording to the method of Mohr f or after reduction with ferrous 
 hydroxide, gave no satisfactory results. The most conven- 
 ient means of reduction we found to be sodium amalgam in 
 alkaline solution. This converted the substance completely 
 into the ferrocyanide, without the separation even of traces of 
 iron. The ferrocyanide was then titrated with potassium 
 permanganate in sulphuric acid solution. The details of the 
 method were first worked out with potassium a-ferricyanide, 
 and the best conditions found to be as follows : The solution 
 of the salt (about 0.2 g.) in 100 c. c. of water, was rendered 
 slightly alkaline with sodium carbonate, and a piece of 5 per 
 cent sodium amalgam was then added. During the reduc- 
 tion the solution was kept on the steam-bath. In the course 
 of an hour or so the mercury was filtered off and the solution 
 cooled and very slightly acidified with sulphuric acid. It was 
 then titrated with a solution of permanganate until the red 
 color of the latter remained for half a minute or more. 
 
 * Zeitschr. anorg. Chem., vi, 174. f Ann. Chem. (Liebig), cv, 62. 
 
POTASSIUM FERRIC YANIDE. 
 
 123 
 
 With potassium a-ferricyanide the following results were 
 obtained : 
 
 1 c. c. KMnO 4 solution = 0.00572 g. Fe = 0.0336 g. 
 K 8 Fe(CN) 6 . 
 
 I . 
 
 II . 
 
 Ill . 
 
 K 8 Fe(CN)e 
 taken. 
 
 0.2596 
 0.2402 
 0.1926 
 
 C. c. KMnO 4 solu- C. c. KMnO 4 
 tion used. calculated. 
 
 7.65 
 7.10 
 5.70 
 
 7.72 
 7.15 
 5.71 
 
 The same method of procedure gave for the new compound 
 the following results : 
 
 I 
 
 II 
 
 III 
 
 IV 
 
 V 
 
 VI 
 
 taken. 
 
 0.1611 
 0.2520 
 0.1946 
 0.2641 
 0.1427 
 0.2208 
 
 C. c. KMnO 4 solu- C. c. KMnO 4 solu- 
 tion used. tion calculated. 
 
 4.55 
 7.20 
 5.40 
 7.35 
 4.10 
 6.30 
 
 4.55 
 7.11 
 5.49 
 7.45 
 4.04 
 6.26 
 
 In order to make sure that in these experiments only potas- 
 sium a-ferricyanide remained in the solutions, or, in other 
 words, that the conversion of the /?-ferricyanide to the latter 
 compound was absolutely quantitative, the amount of 
 a-ferricyanide which the solutions in V or VI contained was 
 gravimetrically determined. This was accomplished by pre- 
 cipitating the titrated solutions with silver nitrate, and esti- 
 mating the percentage from the silver, which was weighed as 
 chloride. The results were very exact, the amount of silver 
 found differing by little more than a milligram from that 
 required by theory : 
 
 V Found, 0.1347 g. Ag = 0.1367 g. K 8 Fe(CN) 8 . 
 
 Calculated, 0.1332 g. Ag = 0.1352 g. K 8 Fe(CN)(,. 
 VI Found, 0.2048 g. Ag = 0.2079 g. K 8 Fe (CN) e . 
 Calculated, 0.2061 g. Ag = 0.2092 g. K 8 Fe(CN) 6 . 
 
 The average percentage of cyanogen in the /3-ferricyanide, 
 
124 
 
 ON AN ISOMER OF 
 
 as calculated from these six determinations, is 45.07. The 
 theory requires 44.96 per cent. 
 
 Potassium /3-ferricyanide, like the normal salt, yields char- 
 acteristic precipitates with the solutions of most of the heavy 
 metals. These have, in general, the same characteristics as 
 the corresponding a-ferricyanides, and in some cases pass over 
 into the latter with extreme ease. The less notable of these 
 precipitates are collected briefly in the table below, in which, 
 for the sake of comparison, the a-ferricyanides are also 
 included. The reactions examined were brought about in 2 
 per cent solutions. In the case of silver, bismuth, stannic 
 tin, and lead, the reactions will be discussed more fully, as 
 they present characteristic points of difference between the 
 two ferricyanic groups. 
 
 /3-Ferricyanides. 
 
 Cd. Dirty green, soluble in 
 (NH 4 ) 2 C0 3 ,HC1, insol- 
 uble in HN0 3 . 
 
 Cu. Yellowish-green, insolu- 
 ble in HN0 8 or HC1. 
 Fe m . Dark yellow coloration, 
 blue precipitate on boil- 
 ing. 
 
 Fe n . Blue precipitate. 
 Hg n . No precipitate. 
 
 Co. Dark red precipitate, in- 
 soluble in HC1 or HN0 3 . 
 
 Mn. Brown precipitate, insol- 
 uble in NH 4 OH, HC1, 
 or HN0 8 . 
 
 Hg 1 . Yellowish-green, floccu- 
 lent precipitate. Rap- 
 idly undergoes re- 
 duction on boiling, 
 becoming blue. 
 
 !Ni. Dark yellowish-green, in- 
 soluble in HNO, or HC1. 
 
 Zn. Yellowish-green, soluble 
 in HC1, (NH 4 ) 2 C0 3 , in- 
 soluble in HNO. 
 
 a-Ferricyanides. 
 
 Pale yellow, soluble in 
 (NH 4 ) 2 C0 3 ,HC1, insol- 
 uble in HN0 3 . 
 
 Dark greenish-yellow, in- 
 soluble in HN0 3 or HC1. 
 
 Dark red coloration, blue 
 precipitate on boiling. 
 
 Blue precipitate. 
 No precipitate. 
 The same. 
 
 The same. 
 
 Pale yellow, undergoes 
 reduction only slowly 
 on boiling. 
 
 Light yellow, same be- 
 havior. 
 Yellow, same behavior. 
 
POTASSIUM FERRICYANIDE. 125 
 
 The /3-ferricyanides of these metals, as is seen, resemble the 
 a-compounds in nearly all particulars, being distinguished 
 from them chiefly in having a more or less pronounced green 
 color. It might be supposed, therefore, that they are identi- 
 cal with them, but colored by slight impurities. But the be- 
 havior of the salts of bismuth, stannic tin, lead, and silver 
 shows that this is not the case. Bismuth a-ferricyanide is a 
 very sparingly soluble, straw-colored precipitate, which is de- 
 posited even from very dilute solutions of potassium ferricy- 
 anide on addition of bismuth nitrate. It is likewise insoluble 
 in concentrated nitric acid. A solution of potassium p-ferricy- 
 anide, on the other hand, when freshly prepared, gives no trace 
 of a precipitate with bismuth nitrate. The solution assumes 
 a slightly greenish tint, but even when concentrated remains 
 otherwise unaltered for some tune. On standing, and espe- 
 cially when exposed to the action of the direct sunlight, it 
 deposits large granules of a black, crystalline compound, 
 which we have not yet fully examined, but which is appar- 
 ently bismuth ferrocyanide. 
 
 Stannic chloride, on the other hand, yields no precipitate 
 with potassium a-ferricyanide, but precipitates the isomer 
 completely. The resulting compound comes down as a slimy, 
 pure green body, which is insoluble in either hydrochloric or 
 nitric acid. We have, however, been unable to obtain it in a 
 state suitable for analysis. 
 
 A comparative study of the lead salts was kindly under- 
 taken by Mr. H. A. North. While lead ferricyanide is not 
 precipitated by potassium a-ferricyanide, under ordinary cir- 
 cumstances, it is much less soluble in water than either the 
 latter compound or lead nitrate. When concentrated solu- 
 tions of the two are mixed in the proper proportions, it slowly 
 separates out in large, dark-red crystals.* Mr. North made 
 this salt by dissolving the calculated quantities, or 3 g. 
 each, of potassium ferricyanide and lead nitrate in 7 c. c. and 
 8 c. c. of water, respectively. On allowing the mixed solutions 
 to stand for a few minutes, he obtained more than a gram of 
 * Rammelsberg, J. prakt. Chem. (2), xxxix, 455. 
 
126 ON AN ISOMER OF 
 
 well-crystallized lead ferricyanide. A similar experiment was 
 then made, with exactly the same quantities, potassium 
 /3-ferricyanide being used instead of the a-salt. No crystalli- 
 zation took place, indicating that lead y3-ferricyanide is much 
 more soluble in water than its isomer. This solution was 
 allowed to stand over night, and by morning, in addition to 
 an efflorescent product, more or less of the normal lead o-ferri- 
 cyanide had crystallized out, the /3-ferricyanide having partially 
 passed over into the latter compound. The solution then gave 
 the usual straw-colored precipitate with bismuth nitrate, with 
 which it had not reacted twelve hours before. 
 
 Various attempts were made to isolate the /3-lead salt by 
 other means, but without much success. The most satisfac- 
 tory results were obtained by dissolving lead oxide in glacial 
 acetic acid, and adding to the solution the calculated quantity 
 of potassium /3-ferricyanide, dissolved in 25 c. c. of glacial acetic 
 acid and 7 c. c. of water. A green, amorphous precipitate 
 separated out, which was readily soluble in water forming a 
 greenish-red solution, and this yielded no precipitate with 
 bismuth nitrate. An analysis of the product showed that it 
 contained 49 per cent lead, and no potassium, but no simple 
 atomic ratio was evident between the lead and nitrogen, 
 which was somewhat low. It is probable, therefore, that the 
 body was a mixture. An experiment made with potassium 
 a-ferricyanide, in a similar manner, yielded a crystalline pre- 
 cipitate of the ordinary lead ferricyanide. 
 
 The most interesting /3-ferricyanide of a heavy metal which 
 we have obtained is the silver salt. This is thrown down 
 quantitatively as a dark brown, flocculent precipitate, which 
 can be readily filtered and washed.* Its most marked charac- 
 teristic is the ease with which it passes into the a-ferricyanide. 
 This takes place simply when the precipitate, suspended in its 
 mother-liquid, is heated to 100. The conversion is indicated 
 by the change in color to the bright orange of the a-ferricy- 
 anide. The silver salt was prepared for analysis by adding the 
 potassium salt to a slight excess of silver nitrate, both being in 
 
 * On drying it forms a light, brown powder. 
 
POTASSIUM FERRICYANIDE. 
 
 127 
 
 Calculated for 
 
 Ag,Fe(CN) 8 . 
 
 60.47 
 
 Found. 
 I. II. 
 
 59.41 60.00 
 
 10.45 
 
 10.84 . . . 
 
 15.67 
 
 15.31 
 
 ice-cold solution. The precipitate was washed with ice-water, 
 then with alcohol, and finally with ether, to secure rapid dry- 
 ing, and then allowed to stand in a vacuum, without exposure 
 to light. The results of the analyses were as follows : 
 
 Ag 
 Ee 
 
 N 
 
 The readiness with which the silver salt passes into silver 
 a-f erricyanide presented another means of ascertaining whether 
 the conversion of the one ferricyanic group into the other is 
 actually quantitative. Weighed quantities of the potassium 
 salt were precipitated with silver nitrate in the cold, and one 
 of the precipitates (I) then rapidly heated in its mother-liquid 
 until it had assumed the orange-red color of silver a-ferricy- 
 anide. The silver in each of the precipitates was then deter- 
 mined, and found to be as calculated for the compound 
 Ag 3 Fe(CN) 6 . 
 
 I 0.2912 g. K 3 Fe(CN) 6 .H 2 gave 0.2729 g. Ag. 
 
 calculated 0.2724 g. Ag. 
 II 0.2470 g. " " gave 0.2291 g. Ag. 
 
 calculated 0.2307 g. Ag. 
 
 The filtrates were then very thoroughly examined for prod- 
 ucts other than potassium nitrate. The filtrate from (I) con- 
 tained a trace of iron, which was precipitated and weighed. 
 It amounted only to 0.15 per cent of the potassium salt. In 
 neither filtrate could either hydrocyanic or cyanic acid be de- 
 tected. The change which the silver salt undergoes on heat- 
 ing must therefore be assumed to take place without the 
 formation of any substance other than silver a-ferricyanide, 
 and to consist simply in a rearrangement of atoms in the mole- 
 cule Ag 3 Fe(CN) 6 . 
 
 It was hoped that by acting on the silver salt with the cal- 
 culated quantity of hydrochloric acid the free /3-ferricyanic 
 acid could be prepared. The solution obtained gave for the 
 
128 ON AN I SOME R OF 
 
 moment no precipitate with bismuth nitrate, showing that the 
 /3-acid had been formed, but the latter passed within a few 
 minutes into the normal acid, and then into further decompo- 
 sition-products. Efforts to prepare the salts of calcium and 
 barium in the same way also failed, the normal ferricyanides 
 being finally obtained. 
 
 In regard to the constitution of the /3-ferricyanic group we 
 have little to state. Any attempt to assign a definite struc- 
 tural formula to it would, for the present, be pure speculation. 
 The suggestion offers itself that one of the two isomers con- 
 tains isonitril groups, the other nitril groups. But this view, 
 at least, is absolutely refuted by the identical behavior of the 
 two on reduction. So far as we can find, there are no cases 
 known where isomers containing respectively the CN and 
 NC groups yield the same product with nascent hydrogen. 
 On the other hand, if both ferricyanides contain only cyano- 
 gen groups, the /3-compound becomes of especial importance 
 because of its bearing on Werner's theory. According to the 
 latter the ferricyanic group is not to be represented by the 
 structural formula, but simply as a radical in which the cyano- 
 gen groups occupy " co-ordination positions " about the ferric 
 atom. Isomerism between two equivalent Fe(CN) 6 groups, 
 if this is so, can be due only to stereochemical causes. The 
 greatest value of Werner's theory lies in the explanation 
 which it offers of the cases of isomerism among the platin- 
 amine and cobaltiamine compounds. The six co-ordinated 
 groups are supposed to occupy the angles of a regular octa- 
 hedron, in the centre of which is the metal. Isomerism is 
 then possible whenever two or more of the co-ordinated groups 
 differ from the others. But according to it a radical in which 
 they are all alike, such as Fe m (CN) 6 , cannot have two differ- 
 ent configurations, or, in other words, exist in isomeric modi- 
 fications. It would seem, therefore, that the existence of 
 potassium /3-ferricyanide stands in direct contradiction to such 
 a theory, at least in its present form. 
 
 It may of course be that the /3-ferricyanide is a mixture 
 of two substances in proportions giving the atomic ratio 
 
POTASSIUM FERRICYANIDE. 129 
 
 K 3 Fe(CN) 6 . But throughout all our investigations we have 
 searched for, and failed to find, a single indication that such 
 is the fact. Its completely crystalline and homogeneous ap- 
 pearance, the constant composition of different products, and 
 all its observed reactions point closely to its being an indi- 
 vidual chemical compound. Our investigations will be con- 
 tinued, however, and we hope to bring more light to bear 
 upon the subject within a short time. 
 
 NEW HAVEN, December 12, 1898. 
 
ON THE FORMATION OF POTASSIUM /9-FERRI- 
 CYANIDE THROUGH THE ACTION OF ACIDS 
 UPON THE NORMAL FERRIC YANIDE.* 
 
 BY JAMES LOCKE AND GASTON H. EDWAKDS. 
 
 WE have shown in a previous paper f that when a solution 
 of potassium ferricyanide is heated for a short time with 
 potassium chlorate and hydrochloric acid, an isomer of this 
 salt is obtained, which crystallizes with one molecule of water. 
 This salt had already been obtained by Skraup J and others, 
 hi an impure state, by the action of oxidizing agents, and was 
 regarded by them as an oxidation product of the normal ferri- 
 cyanide, with the formula K 2 Fe(CN) 6 . At the time of our 
 first investigation we likewise ascribed its formation to the 
 oxidizing action of the potassium chlorate, but have since 
 found that it is due to the hydrochloric acid alone, and is, in 
 fact, formed more or less readily by the action of any acid on 
 potassium ferricyanide. Its preparation by former experi- 
 menters can thus be readily explained, as their operations 
 were all carried on in acid solutions. 
 
 In our first experiments to investigate this point we used 
 the same quantity of hydrochloric acid as in the preparation 
 of the salt with potassium chlorate, that is, the amount re- 
 quired according to the equation given by Skraup, 
 
 6K 8 Fe(CN) 6 + KC10 8 + 6HC1 = 6K 2 Fe(CN) 6 + 7KC1 + 3H 2 0, 
 
 and otherwise performed the experiments exactly as described 
 in our previous article. 50 g. of potassium ferricyanide 
 were dissolved in 100 c. c. of water, heated to boiling, and 18 
 c. c. of concentrated hydrochloric acid (sp. gr. 1.19), diluted 
 
 * Amer. Chem. Jour., xxi, May, 1895. 
 
 t Ibid., 193. } Ann. Chem. (Liebig), clxxxix, 368. 
 
FIG. I. Potassium /3-ferricyanide. 
 
 FIG- II. Potassium a-ferricyanide. 
 
FORMATION OF POTASSIUM p-FERRICYANIDE. 131 
 
 with three times their volume of water, were added, and the 
 mixture allowed to stand on the water-bath. Small portions 
 were taken out at short intervals, cooled, precipitated with 
 equal volumes of 95 per cent alcohol, and the precipitates 
 filtered off. These various preparations were then tested with 
 bismuth nitrate, with which the a-ferricyanide gives a yellow 
 precipitate, and the /3-ferricyanide none. Subsequently the 
 percentages of cyanogen which they contained were determined 
 by reduction with sodium amalgam, and titration to potassium 
 a-ferricyanide with potassium permanganate.* 
 
 Two samples of the solution, taken out after the reaction 
 had proceeded one and one-half minutes and three minutes, 
 respectively, gave with alcohol yellow and yellowish-green 
 precipitates, the colors of which were due to the presence of 
 unchanged a-ferricyanide. These showed, with bismuth 
 nitrate, the test characteristic of the latter. At the end of five 
 minutes the originally reddish-yellow solution had assumed the 
 peculiar red-violet color of the /3-salt, and the precipitated 
 samples then consisted of a completely homogeneous mass of 
 well-formed, olive-green crystals. These showed all the 
 properties of potassium /3-ferricyanide, K 8 Fe(CN) 6 .H 2 O, as 
 previously described by us. They are most readily recognized 
 by their very characteristic crystal-habit, showing the forms 
 illustrated in Fig. 1. This is taken from a photomicrograph, as 
 is also Fig. 2, which shows crystals of the normal ferricyanide 
 obtained under similar conditions. At the same time the per- 
 centage of cyanogen rapidly fell to a value corresponding 
 closely to that calculated for the yS-salt; that is, 44.96 per 
 cent. The sample obtained after the reaction had proceeded 
 for ten minutes was of a greenish-black color, and showed no 
 distinct crystallization, indicating that further decomposition 
 had taken place. In this, and subsequent samples also, the 
 percentage of cyanogen had fallen considerably below the 
 value for the /3-compound. 
 
 The analytical results of this series of experiments are 
 tabulated below : 
 
 * Amer. Chem. Jour., xxi, 193. 
 
132 
 
 ON THE FORMATION OF 
 
 Series A. 
 
 50 g. K 8 Fe(CN) 6 ; 18 c. c. cone. HC1. 
 
 1 c. c. KMn0 4 solution = 0.03704 g. K 8 Fe(CN) 6 . 
 
 No. 
 
 Time. 
 
 Color. 
 
 M 
 
 d 3 
 
 C) | 
 
 B 
 
 0.2 
 
 
 Min. 
 
 
 * 
 
 
 
 
 
 I 
 
 n 
 
 yellow 
 
 0.1881 
 
 4.9 
 
 4.81 
 
 5.08 
 
 45.74 
 
 II 
 
 3 
 
 yellowish-green 
 
 0.2616 
 
 6.8 
 
 6.70 
 
 7.06 
 
 45.65 
 
 III 
 
 5 
 
 olive 
 
 0.2073 
 
 5.3 
 
 5.31 
 
 5.59 
 
 44.89 
 
 IV 
 
 7 
 
 olive 
 
 0.2919 
 
 7.4 
 
 7.47 
 
 7.88 
 
 44.52 
 
 V 
 
 10 
 
 green-black 
 
 0.2376 
 
 5.9 
 
 6.08 
 
 6.42 
 
 43.60 
 
 VI 
 
 15 
 
 green-black 
 
 
 
 
 
 
 VII 
 
 20 
 
 green-black 
 
 0.3014 
 
 7.3 
 
 7.71 
 
 8.17 
 
 42.53 
 
 Cyanogen calculated for K 3 Fe(CN) 6 , 47.41 per cent. 
 
 " " K 8 Fe(C:N") 6 .H 2 0, 44.96 per cent. 
 
 The question next suggested itself, whether it was neces- 
 sary to have hydrochloric acid present in definite molecular 
 quantity, as above, or whether its action was not merely one 
 of catalysis, and caused by the presence even of small quanti- 
 ties. In order to decide this, experiments similar to the above 
 were made, hi which one-half, one-fourth, and one-eighth of the 
 amount of the acid there used was employed. It was found, 
 not only that the /3-compound was formed in each case, but 
 that its subsequent decomposition by the acid takes place 
 much more slowly than when the acid is more concentrated. 
 The reactions could, therefore, be much more easily followed 
 in these experiments, as a considerable interval elapsed between 
 the point at which the a-salt disappeared and that at which 
 the /3-salt noticeably began to decompose. The results of 
 these experiments are tabulated below: 
 
POTASSIUM p-FERRICYANIDE. 
 
 133 
 
 Series B. 
 50 g. K 8 Fe(CN) 8 ; 9 c. c. cone. HCi. 
 
 it 
 
 83 
 
 No. Time. Color. |to 
 
 Min. g. 
 
 I 5 yellowish-green 0.2152 
 
 II 10 olive 0.2222 
 
 III 15 olive 0.2233 
 
 IV 20 green-black 0.2155 
 V 25 green-black 0.2393 
 
 Series C. 
 
 50 g. K 8 Fe(CN) 8 ; 4.5 c. c. cone. HCI. 
 
 fife 
 
 i&A 
 
 II 
 
 5.6 
 
 5.51 
 
 5.81 
 
 45.74 
 
 5.7 
 
 5.68 
 
 5.90 
 
 45.03 
 
 5.7 
 
 5.72 
 
 6.03 
 
 44.82 
 
 5.2 
 
 5.52 
 
 5.82 
 
 42.37 
 
 5.7 
 
 6.12 
 
 6.49 
 
 41.83 
 
 I 
 
 2 
 
 yellow 
 
 0.3535 
 
 9.3 
 
 9.05 
 
 9.54 
 
 46.19 
 
 II 
 
 4 
 
 yellowish-green 
 
 0.3421 
 
 8.9 
 
 8.75 
 
 9.24 
 
 45.68 
 
 III 
 
 6 
 
 olive 
 
 0.1901 
 
 4.9 
 
 4.86 
 
 5.13 
 
 45.26 
 
 IV 
 
 8 
 
 olive 
 
 0.2569 
 
 6.6 
 
 6.57 
 
 6.94 
 
 45.11 
 
 V 
 
 10 
 
 olive 
 
 0.2263 
 
 5.8 
 
 5.79 
 
 6.11 
 
 45.00 
 
 VI 
 
 15 
 
 olive 
 
 0.2223 
 
 5.7 
 
 5.69 
 
 5.90 
 
 45.03 
 
 VII 
 
 20 
 
 green 
 
 0.1994 
 
 5.0 
 
 5.10 
 
 5.38 
 
 44.03 
 
 VIII 
 
 25 
 
 green-black 
 
 0.3195 
 
 8.0 
 
 8.18 
 
 8.63 
 
 44.00 
 
 IX 
 
 30 
 
 green-black 
 
 0.2198 
 
 5.5 
 
 5.63 
 
 5.93 
 
 43.94 
 
 In Series D it will be seen that in the first fifteen minutes 
 the percentage of cyanogen fell from the value calculated for 
 the a-salt, or 47.41 per cent, to 45.02 per cent, or nearly the 
 value calculated for the yS-salt, a loss of 2.4 per cent. From 
 this point on, the precipitated samples gave no reactions with 
 bismuth nitrate. During the next twenty minutes the per- 
 centage of cyanogen fell only 0.45 per cent, and the fractions 
 taken out in this interval were all thoroughly homogeneous 
 and well crystallized. Beyond this point the decomposition 
 again became more rapid, and the samples obtained were of 
 far less satisfactory appearance than those in numbers III-VI. 
 
134 
 
 ON THE FORMATION OF 
 
 Series D. 
 50 g. K,Fe(CN), ; 2.25 c. c. cone. HC1. 
 
 la ^ 
 
 S&A 
 
 111 
 
 No. 
 
 II 
 III 
 
 IV 
 
 V 
 
 VI 
 
 VII 
 
 VIII 
 
 IX 
 
 X 
 
 XI 
 XII 
 
 Time. 
 Min. 
 
 5 
 
 10 
 15 
 20 
 25 
 35 
 45 
 55 
 65 
 
 Hours. 
 
 1* 
 
 
 
 Color. 
 
 yellow 
 
 yellow 
 
 yellowish-green 
 
 olive 
 
 olive 
 
 olive 
 
 olive 
 
 green 
 
 green-black 
 
 green-black 
 green-black 
 green-black 
 
 ! j 
 
 H 
 
 ti 
 
 
 
 M|I 
 
 I! 
 
 0.2256 
 
 5.9 
 
 5.77 
 
 6.06 
 
 45.92 
 
 0.1510 
 
 3.9 
 
 3.86 
 
 4.08 
 
 45.02 
 
 0.2121 
 
 5.4 
 
 5.43 
 
 5.73 
 
 44.71 
 
 0.2584 
 
 6.6 
 
 6.61 
 
 6.98 
 
 44.85 
 
 0.2639 
 
 6.7 
 
 6.75 
 
 7.12 
 
 44.57 
 
 0.2113 
 
 5.3 
 
 5.41 
 
 5.70 
 
 44.04 
 
 0.2036 
 
 5.0 
 
 5.21 
 
 5.50 
 
 43.12 
 
 0.2171 
 
 5.4 
 
 5.55 
 
 5.86 
 
 43.67 
 
 0.1745 
 
 4.2 
 
 4.46 
 
 4.71 
 
 42.27 
 
 0.2015 
 
 4.9 
 
 5.17 
 
 5.42 
 
 42.70 
 
 0.2010 
 
 4.5 
 
 5.14 
 
 5.43 
 
 39.31 
 
 In Series B and similar results were obtained, but the re- 
 actions proceeded much more rapidly. In (7, for instance, the 
 samples ceased to react with bismuth nitrate in eight minutes, 
 and the #-salt was noticeably decomposed at the end of 
 twenty minutes. It must, therefore, be concluded that the 
 velocity of the reaction is directly dependent upon the concen- 
 tration of the acid. 
 
 Similar results were obtained when other acids were substi- 
 tuted for hydrochloric acid. Experiments were instituted 
 with sulphuric, oxalic, and acetic acids. No quantitative de- 
 terminations were made, however, as it was found impossible, 
 in the first two cases, to obtain the precipitates free from 
 potassium sulphate and potassium oxalate. But that the re- 
 action proceeded in the same way as when hydrochloric acid 
 was used, was readily recognized by the color of the solution 
 and by the appearance of the precipitates. With oxalic acid 
 the /3-salt was formed less readily than with sulphuric acid, 
 
POTASSIUM fr-FERRICYANIDE. 135 
 
 though in both cases the conversion of the a-salt into its iso- 
 mer took place more slowly than it did in hydrochloric acid 
 solution. It would thus appear that the catalytic action of 
 different acids, to which the formation of the /3-salt is due, is 
 directly dependent upon the degree to which the acid used 
 undergoes ionization. According to this, an acid such as 
 acetic, which ionizes but slightly, should have only a very 
 slow action. This was found to be the case, for with dilute 
 acetic acid we could obtain no /3-salt at all, and a solution of 
 ferricyanide, heated for an hour with the concentrated acid 
 (2 molecules), still gave a slight precipitate with bismuth 
 nitrate. 
 
 NEW HAVEN, February, 1899. 
 
ON THE SEPARATION OF TUNGSTIC AND 
 SILICIC ACIDS. 
 
 BY H. L. WELLS AND F. J. METZGER. 
 
 IN a recent number of a German periodical* appears an 
 article by Otto Herting of Philadelphia in which the assertion 
 is made that the method given in the text-books for expelling 
 silica from tungstic acid by means of hydrofluoric acid is incor- 
 rect. This statement is made on the ground of alleged numer- 
 ous quantitative experiments with mixtures of pure tungstic 
 acid and pure ignited silicic acid, but no details in regard to 
 the results are given. Herting believes that upon ignition 
 silicic and tungstic acids form a silico-tungstic acid which is 
 volatile when treated with hydrofluoric acid, and finally says 
 that he should be pleased if by means of his article he should 
 bring about the more careful study of the " action of hydro- 
 fluoric acid upon tungstic acid in the presence of silicic acid." 
 
 Since Herting's statement throws doubt upon a method that 
 is generally used, we have undertaken an examination of the 
 matter. For this purpose we dissolved some of Kahlbaum's 
 tungstic acid in ammonia, precipitated with nitric acid, washed 
 with water by decantation, digested repeatedly with sulphuric 
 acid of sp. gr. 1.378 to separate any molybdic acid that might 
 possibly be present,! washed the residue and ignited it. The 
 tungstic acid thus prepared was used for the experiments that 
 follow. 
 
 A weighed quantity of tungstic acid in a platinum crucible 
 was mixed with about an equal quantity of pure silica. The 
 mixture was covered with dilute sulphuric acid, a liberal 
 amount of pure hydrofluoric acid was added, the liquid was 
 carefully evaporated and the residue was ignited over a Bun- 
 
 * Zeitschr. f. angew. Chem., 1901, 165. 
 
 t See Ruegenberger and Smith, Jour. Amer. Chem. Soc., xxii, 772. 
 
SEPARATION OF TUNGSTIC AND SILICIC ACIDS. 137 
 
 sen burner. Then another portion of silica was added and the 
 operation was repeated. The results are shown in the follow- 
 ing table : 
 
 Taken. WO 3 found after 
 
 WO, SiO, 1st operation. 2d operation. 
 
 I 1928 .2 .1927 .1928 
 
 II 2097 .2 .2097 .2096 
 
 III 2100 .2 .2099 .2100 
 
 IV 1999 .2 .2000 .1998 
 
 The greatest error found in these experiments is .0001 g., 
 and they show that the process is perfectly exact under these 
 conditions. 
 
 Other experiments showed that long ignition of the mixed 
 tungstic and silicic acids over the Bunsen burner before expel- 
 ling the silicic acid had absolutely no effect upon the results. 
 
 It was thought that in the absence of sulphuric acid a loss 
 might occur by the treatment of tungstic acid with hydro- 
 fluoric acid, and the following experiments were made to test 
 this point, no sulphuric acid being used : 
 
 Taken. Found. 
 
 WO, SiO, WO 8 
 
 g- g- g- 
 
 .1983 .2 .1983 
 
 .2102 .2 .2100 
 
 .2106 .2 .2105 
 
 .1996 .2 .1994 
 
 In these experiments the greatest error, .0002 g., is well 
 within reasonable limits ; hence it is evident that the absence 
 of sulphuric acid has no effect. It is to be noticed that in 
 these cases also the tungstic acid was ignited by the Bunsen 
 flame only. 
 
 Attention should be called to the fact that tungstic acid 
 must not be ignited by means of the blast-lamp, since at the 
 temperature thus produced it volatilizes to a considerable ex- 
 tent. The books of reference do not give proper warning in 
 regard to this matter. The following table gives the results 
 of a series of experiments made by heating over the blast-lamp 
 
138 SEPARATION OF TUNGSTIC AND SILICIC ACIDS. 
 
 in a platinum crucible some of the substance, which showed no 
 loss of weight over the Bunsen burner. 
 
 Weight of WO 8 . Loss. 
 
 Taken 3007 
 
 Ignited for 2 minutes 2978 .0029 
 
 " " again . . . .2962 .0016 
 
 " "... .2946 .0016 
 
 " " "... .2932 .0014 
 
 " "... .2924 .0008 
 
 " " "... .2916 .0008 
 
 " " "... .2906 .0010 
 
 " 5 " "... .2872 .0034 
 
 Total loss, .0135 
 
 All the ignitions except the last were made with a lamp pro- 
 vided with a water-blast, which gave a flame of only moderate 
 power. The last ignition was made with a lamp connected 
 with a foot-bellows, which gave a considerably higher tempera- 
 ture. It is noticeable that the losses show a tendency to di- 
 minish after the first ignition, but this is probably due to a 
 change in the physical condition of the oxide rather than to 
 the removal of some more volatile substance. It is hardly 
 possible that our carefully purified tungstic acid, which showed 
 no loss when heated with a good Bunsen burner, could contain 
 an amount of molybdic acid or other volatile substance suffi- 
 cient to give the results that have been obtained. The loss 
 shown in the table above amounts to nearly 5 per cent, while 
 in another experiment .1955 g. of tungstic acid lost over 7 per 
 cent after heating with the blast-lamp for twenty minutes. It 
 should be stated that the platinum crucible in which these 
 ignitions were made showed no loss in weight after it had 
 been cleaned. 
 
 We have shown that Herting's criticism of the usual method 
 for separating silicic and tungstic acids is without foundation, 
 and it appears probable that his difficulties were due to ignite 
 ing tungstic acid at a too elevated temperature. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1901. 
 
ON A SALT OF QUADRIVALENT ANTIMONY. 
 
 BY H. L. WELLS AND F. J. METZGER. 
 
 WHILE engaged in purifying some caesium material, we 
 precipitated a large quantity of the salt Cs 3 Sb 2 Cl 9 by adding 
 an excess of antimony trichloride to a hydrochloric acid solu- 
 tion of impure caesium chloride. A small amount of caesium 
 remained in the filtrate, and, wishing to recover this, we added 
 some lead nitrate solution, and passed chlorine gas into the 
 liquid in order to precipitate the very insoluble lead tetra- 
 chloride salt, Cs 2 PbCl 6 .* Much to our surprise, the pre- 
 cipitate, while showing the usual octohedral form, was bright 
 green hi color, whereas the pure lead salt is bright yellow. 
 The product was found to contain a small quantity of anti- 
 mony, and by preparing the lead salt after the addition of 
 varying quantities of antimony trichloride we obtained prod- 
 ucts of various colors, from yellowish green to dark bluish 
 green. The following determinations were made in a lighter 
 and a darker product : 
 
 Light green. Dark green. 
 
 Antimony . . . 0.92 per cent 1.44 per cent 
 
 These results indicated that antimony was not an essential 
 constituent of the compound, and it seemed probable that 
 some salt of antimony isomorphous with Cs 2 PbCl had crys- 
 tallized with it. We drew the further inference that the 
 isomorphous antimony compound must be strongly colored, 
 probably blue. Fortunately we had access to the reprint of 
 an article by Setterberg f (which, as far as we can find, has 
 not been noticed in any of the books of reference) where a 
 peculiar black caesium-antimony salt is described. Setterberg 
 
 * Amer. Jour. Sci. (3), xlvi, 180. 
 
 t Ofversigt K. Vetensk.-Akad. Forhandl., 1882, 23. 
 
140 ON A SALT OF 
 
 made this compound by boiling a solution of antimony tri- 
 chloride in strong hydrochloric acid with antimony penta- 
 chloride and caesium chloride in excess. He ascribes to it 
 the formula 2CsCLSbCl4 or 4CsCl.SbCl 8 .SbCl 6 , and states 
 that it forms black, very small, short prisms. It was evi- 
 dent that a salt of this composition might be expected to 
 crystallize with 2CsCl.PbCl 4 , and if so, the fact would be a 
 strong argument in favor of Setterberg's first formula, but 
 his description of the form of the salt gave no evidence of 
 the isomorphism of the two compounds. 
 
 We have prepared Setterberg's salt under varying con- 
 ditions, and have confirmed his formula, as is shown by the 
 following analysis of a very pure product : 
 
 Pound. 
 
 Caesium ..... 44.92 44.41 
 
 Antimony ..... 20.23 20.03 
 
 Chlorine ..... 35.13 35.56 
 
 100.28 100.00 
 
 As far as can be judged by careful microscopic examination, 
 this salt crystallizes in perfect octahedra. We could find no 
 indication of a prismatic form corresponding to Setterberg's 
 description ; hence it is probable that his crystals were so small 
 that he did not make them out clearly. No optical exami- 
 nation of the crystals could be made on account of their 
 complete opacity. The color of the crystallized substance is 
 absolutely black, but when a little of it is rubbed between the 
 ground surfaces of a glass-stoppered bottle, it shows a very 
 strong, dark blue color. 
 
 From the' crystalline form and color of this curious salt 
 there can be no doubt that it was the substance which gave 
 the green color to our products of Cs 2 PbCl 6 , and it seems 
 certain that the two salts are isomorphous. The color of the 
 antimony salt indicates that it is not a compound containing 
 antimony trichloride and pentachloride, for the known cae- 
 sium double salts with these chlorides, 3CsC1.2SbCl 8 and 
 
QUADRIVALENT ANTIMONY. 141 
 
 CsCl.SbCls,* are colorless or nearly so, We are convinced, 
 therefore, that this black salt, Cs 2 SbCl, is a member of the 
 well-known group of octahedral double halides of quadriva- 
 lent elements, among which are K 2 PtCl 6 , Cs 2 SrCl, Cs 2 TeCl 6 , 
 Cs 2 TeI 6 , K 2 PbCl 6 , and Cs 2 PbCl 8 , and that it is a double salt 
 of antimony tetrachloride, SbCl 4 . 
 
 Judging from the color of the double salt, the tetrachloride 
 SbCl 4 must be black. This color would be entirely unex- 
 pected, since SbCl s and SbCl 5 are colorless. We have tried 
 in vain to get some evidence of the separate existence of a 
 black chloride by making mixtures of the two known chlorides 
 at low and high temperatures and by treating the salt CsaSbCle 
 with cold concentrated sulphuric acid. 
 
 The oxide of antimony Sb 2 O 4 , or SbO 2 , is usually regarded 
 as a compound of Sb 2 O 8 and Sb 2 O 6 . It is possible that this 
 may be a true dioxide, but it would be expected that an oxide 
 which corresponded to a black chloride would be black also, 
 for, if there is a difference in color, oxides are usually darker 
 than the corresponding chlorides. 
 
 We have prepared a jet-black double bromide by using a 
 method exactly similar to that by means of which the bkck 
 double chloride was made. Judging from analogy, this is 
 probably Cs 2 SbBr 6 , but we were unable to get a product that 
 was pure enough for analysis, as it was always mixed with 
 more or less light^colored impurity. We attempted, without 
 success, to make the corresponding iodide and fluoride. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1901. 
 
 * Setterberg, loc. cit. 
 
ON THE PURIFICATION OF CJESIUM MATERIAL. 
 
 BY H. L. WELLS. 
 
 SEVERAL years ago * I recommended the use of the yellow 
 salt 2CsCl.PbCl 4 as a means of precipitating caesium from its 
 solutions. Subsequent experience has shown that the method 
 is very convenient for removing small quantities of the rare 
 metal from all sorts of solutions, particularly from those from 
 which the greater part of it has been precipitated by other 
 means. For large quantities of caesium, however, the lead 
 tetrachloride method, used alone, is inconvenient on account 
 of the large amount of chlorine that must be used, and because 
 it is possible to obtain only very dilute solutions of lead chlo- 
 ride or of the white double salt CsC1.2PbCl 2 .t 
 
 I have used extensively the salt just mentioned, as a means 
 of precipitating caesium from concentrated solutions of the 
 crude extract of pollucite made by means of hydrochloric acid. 
 The precipitation is made by adding a hot, concentrated solu- 
 tion of the calculated amount of lead nitrate. The precipitate 
 is crystalline and can be readily washed with dilute hydro- 
 chloric acid, then it can be very easily decomposed by boiling 
 with ammonium carbonate solution. The method is satisfac- 
 tory in a case where very little potassium and practically no 
 rubidium are present, for although it does not give a good 
 separation from these metals, the precipitation of caesium is 
 nearly complete. The precipitate CsC1.2PbCl 2 , however, is 
 very heavy compared with the amount of caesium that it con- 
 tains, and its decomposition with ammonium carbonate pro- 
 duces four molecules of ammonium chloride for one of caesium 
 chloride. For these reasons the method has been abandoned 
 in this laboratory in favor of the method of Godeffroy. 
 
 * Amer. Jour. Sci. (3), xlvi, 186. f Ibid. (3), xlv, 121. 
 
PURIFICATION OF CAESIUM MATERIAL. 143 
 
 In precipitating caesium according to GodefTroy's method 
 I am accustomed to use less concentrated hydrochloric acid 
 solutions than those that have been recommended. About one- 
 third or one-half of the total volume of concentrated hydro- 
 chloric acid answers very well, and a solution of this kind 
 possesses the advantage that it can be filtered by means of 
 paper. To such a solution antimony trichloride, also dissolved 
 in hydrochloric acid, is added as long as a precipitate forms. 
 The latter is collected, best on a Biichner filter, and well washed 
 with cold dilute hydrochloric acid. 
 
 To the filtrate and washings from the antimony salt, with- 
 out concentrating them, lead nitrate is added at the rate of 2 
 or 3 g. per liter. This salt should be dissolved in water before 
 it is added. Chlorine gas is then passed in until the solution 
 is saturated with it. After standing for a few hours the liquid 
 is decanted from the precipitate of Cs 2 PbCl 6 (which is usually 
 colored green from the presence in it of the dark blue salt 
 Cs 2 SbCl 6 *) and is tested by the addition of a little more lead 
 nitrate, and chlorine also if necessary. This precipitation is 
 so complete, as far as caesium is concerned, that the liquid may 
 be finally discarded. A liter of such a liquid will probably 
 hold in solution only about 0.1 g. of Cs 2 PbCl 6 , but the rubid- 
 ium salt is nearly 100 times more soluble. The lead tetra- 
 chloride salt should be washed on a Biichner funnel with cold, 
 dilute hydrochloric acid. 
 
 Formerly I was accustomed to decompose the antimony 
 salt 3CsC1.2SbCl 8 f by suspending it in water and pass- 
 ing in hydrogen sulphide gas, but for large quantities this is 
 a very slow and laborious operation, and it is much more 
 convenient to treat it in a large porcelain dish with boiling 
 dilute ammonium hydroxide. The antimonious oxide thus 
 produced is dense and easy to filter and wash ; moreover it 
 
 * Concerning this compound, see the preceding article. 
 
 t The composition of this compound was very incorrectly determined by 
 Godeffroy, who gave it the formula 6CsCl.SbCl 8 . Setterberg first arrived 
 at the correct formula, which was afterwards confirmed by Remsen and 
 SaunderS; and also by Muthmann, and by Wells and Foote. 
 
144 ON THE PURIFICATION OF 
 
 may be readily dissolved in hydrochloric acid and used again 
 for precipitating caesium. A little antimony goes into solu- 
 tion in the ammonia, but this is readily removed, best after the 
 liquid has become slightly acid by evaporation, by passing in 
 hydrogen sulphide and filtering. 
 
 The removal of ammonium chloride on a large scale by 
 ignition is a very slow operation. It is therefore best, in 
 most cases, after the hydrogen sulphide has been removed by 
 evaporation, to add to the concentrated solution of caesium 
 and ammonium chlorides, in a large porcelain dish, a liberal 
 amount of concentrated nitric acid and to heat cautiously 
 until frothing has ceased. When further additions of strong 
 nitric acid produce no change of color or evolution of gas 
 upon heating, the ammonium salts are entirely destroyed and 
 the caesium chloride changed to nitrate.* The solution of 
 caesium nitrate in nitric acid is then evaporated to dryness, 
 and the residue is heated until the acid nitrate f is decom- 
 posed, and the normal nitrate is more or less completely 
 fused. The evaporation to dryness of the nitrate in the 
 presence of nitric acid over a free flame in a dish is not a 
 difficult operation, since decrepitation takes place to a much 
 less extent than is the case when the chloride is evaporated 
 in the same way. A large mass of fused caesium nitrate 
 should not be allowed to solidify quietly in the bottom of a 
 porcelain dish, for on cooling it is liable to break the dish by 
 its contraction. If it is stirred as it solidifies, this danger is 
 avoided. 
 
 Caesium nitrate is one of the most beautiful of salts for 
 recrystallization. Like potassium nitrate, it is much more 
 soluble in hot water than in cold, and it crystallizes upon 
 slowly cooling a hot concentrated solution in the form of large 
 colorless prisms. I have found that recrystallizing this salt 
 tends to purify it from traces of rubidium, for a large quantity 
 
 * This method of destroying ammonium chloride is due to J. Lawrence 
 Smith. See Amer. Jour. Sci. (2), xxxiv, 367. 
 
 t A description of the acid caesium nitrates will be published soon from 
 this laboratory. 
 
CAESIUM MATERIAL. 145 
 
 of the salt which showed no rubidium spectrum, upon sys- 
 tematic recrystallization showed a distinct spectrum for that 
 metal in the final mother-liquor. This, however, is a slow 
 method of purification if it is desired to carry the latter to the 
 last degree, but the nitrate obtained by one or two recrystalli- 
 zations is pure enough for any ordinary purpose, unless the 
 original antimony salt was contaminated to an unusual extent. 
 
 The nitrate is readily converted into carbonate by mixing it 
 with two parts of pure oxalic acid, adding a little water, 
 evaporating to dryness and fusing the residue in a platinum 
 crucible.* 
 
 Where the highest degree of purity in a caesium salt is 
 desired, the salt CsCl 2 I t by its recrystallization probably 
 offers one of the best means for accomplishing that object. 
 It is enormously less soluble than the corresponding rubidium, 
 potassium, sodium, and lithium salts. I have found that this 
 salt may be made very conveniently from the nitrate by dis- 
 solving one part of the latter together with one atomic pro- 
 portion of iodine in about ten parts of 1 : 1 hydrochloric acid 
 at a temperature just below boiling. The solution takes 
 place quickly, and, upon cooling, the beautiful yellow salt 
 crystallizes out. It is recrystallized by solution in hot 1 : 1 
 hydrochloric acid and cooling. One or two recrystallizations, 
 when each crop is well drained and washed with a little cold 
 dilute hydrochloric acid, give, even from impure materials, a 
 remarkably pure product. From this trihalide by ignition, 
 best at a very gentle heat so that the mass does not fuse, 
 pure caesium chloride may be prepared. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1901. 
 
 * Method of J. Lawrence Smith, loc. cit. 
 t Amer. Jour. Sci. (3), xliii, 17. 
 
 10 
 
ON THE ACID NITRATES. 
 
 BY H. L. WELLS AND F. J. METZGER. 
 
 A NUMBER of years ago, it was noticed by one of us that when 
 caesium nitrate is evaporated with strong nitric acid, the last 
 part of the acid is expelled with some difficulty, and the 
 probable existence of a rather stable acid nitrate suggested 
 itself. Mr. A. P. Beardsley, of this laboratory, undertook an 
 examination of the matter last year, and, upon dissolving 
 caesium nitrate in hot nitric acid of sp. gr. 1.42 and cooling, 
 obtained beautifully crystallized crops of the salt HNO 8 . 
 CsNO 3 . The circumstances were such that Mr. Beardsley's 
 work in this direction was interrupted, and we have under- 
 taken a further examination of the subject. Ditte* has de- 
 scribed the following list of acid nitrates : 
 
 KN0 8 .2HN0 8 
 
 NH 4 N0 3 .HN0 8 
 
 NH 4 N0 3 .2HN0 8 
 
 [2RbN0 3 .5HN0 8 ] 
 
 [T1N0 8 .3HN0 8 ] 
 
 The two formulae enclosed in brackets do not correspond 
 to the others nor to the caesium salt mentioned above, and we 
 have found that they are incorrect. In making these alleged 
 compounds, 2RbNO 8 .5HNO 8 and T1NO 8 .3HNO 8 , Ditte satu- 
 rated " monohydrated " nitric acid (the liquid corresponding 
 to the formula HNO 8 ) with the normal nitrates and then 
 simply analyzed the solution. He had prepared the other acid 
 nitrates in the same way, but in these cases, having performed 
 the operation above their melting-points, he was able to make 
 definite compounds, and showed that the solutions became 
 perfectly solid at definite temperatures. 
 
 * Ann. de Chim. (5), xviii, 320 (1878). 
 
ON THE ACID NITRATES. 147 
 
 We have easily succeeded in crystallizing two caesium and 
 two rubidium acid nitrates, and a single thallium salt, and find 
 that the salts of the last two metals do not correspond to the 
 formulae given by Ditte. We have confirmed the composition 
 of his potassium acid nitrate, and since his ammonium salts 
 correspond exactly to the caesium and rubidium compounds, 
 we assume that they are correct. Our revision and additions 
 lead to the following list of acid nitrates : 
 
 KN0 8 .2HN0 8 
 
 KbN0 8 .HN0 8 KbN0 8 .2HN0 8 
 
 CsN0 8 .HN0 8 CsISr0 8 .2HNOg 
 
 ..... T1N0 8 .2HNO, 
 
 These salts belong to two types, of which the diacid form is 
 evidently most readily produced, since thallium and potassium 
 apparently do not yield the other. The caesium salts are 
 more easily prepared and are more stable than the others. 
 
 It has been supposed that " monobasic " acids are incapable 
 of forming acid salts, and it is evident that this belief prevails 
 at the present tune, for the following statement is made in a 
 recently published text-book : " Acids like hydrochloric and 
 nitric acids have not the power to form acid salts. They 
 are called monobasic acids." This statement seems to be 
 too sweeping, for there are many acid chlorides, some 
 of which are well known and remarkably stable; in fact, 
 their stability when in solution apparently leads to the 
 opinion that they are not real acid salts. The following list 
 will serve as examples of these compounds: 2HCl.PtCl 4 ; 
 HCl.AnCl 8 ; 2HCl.SnCl 4 ; HCl.ZnCl 2 .2H O ; 2HCl.CuCl ; 
 2HCl.CuCl a ; 5HCl.SbCl 5 .10H 2 O. Iodides and bromides are 
 also known, for example, 2HBr.SnBr 4 and HI.Agl. The ease 
 with which hydrofluoric acid forms acid salts is remarkable, and 
 many acid salts of univalent organic acids, such as acetic, 
 propionic, and valerianic acids are well-known. Since it is now 
 shown that the acid nitrates are well defined and in some cases 
 comparatively stable bodies, the old idea that only " polybasic " 
 
148 ON THE ACID NITRATES. 
 
 acids are capable of forming acid salts should be entirely 
 abandoned. The circumstance that acid nitrates are not more 
 numerous is doubtless connected with the fact that double 
 nitrates in general are rare. 
 
 Method of Preparation. The monoacid salts RbNO 3 .HNO 3 
 and CsNO 8 .HNO 8 are readily prepared by saturating nitric 
 acid of sp. gr. 1.42 with the normal nitrates at a gentle heat 
 and cooling to crystallization. The use of a freezing mixture, 
 such as ice and salt, is desirable in making the rubidium com- 
 pound, but it is not necessary for the caesium salt. Only 
 normal potassium nitrate and thallous nitrate could be crystal- 
 lized from nitric acid of this strength. 
 
 The diacid salts KNO 3 .2HNO 3 , RbN0 8 .2HNO 3 , CsNO 3 . 
 2HNO 8 , and T1NO 3 .2HNO 8 were all prepared by dis- 
 solving the normal nitrates to saturation in nitric acid of 
 sp. gr. 1.50 and cooling with a freezing mixture, usually con- 
 siderably below 0. In preparing the thallous compound 
 warming must be avoided, because it causes the oxidation of a 
 part of the thallium to thallic nitrate. When this has hap- 
 pened, we find that the addition of alcohol, after the removal 
 of most of the strong nitric acid by evaporation, is a convenient 
 means of converting thallic nitrate into the thallous salt. 
 
 Properties. The monoacid salts RbNO 8 .HNO 8 and 
 CsNO 3 .HNO 3 form large, flat masses of small, colorless, trans- 
 parent crystals, apparently octahedra, arranged in parallel po- 
 sition. The diacid salts RbNO 8 .2HNO 8 and T1NO 3 .2HNO 8 
 form beautiful, colorless, transparent needles, while CsNO 8 . 
 2HNO 8 forms large, thin, colorless, transparent plates. 
 
 All the acid nitrates give off nitric acid more or less rapidly 
 upon exposure to the air at ordinary temperature. The salt 
 T1NO 8 .2HNO 8 must be dried on paper in a very cool place, 
 as its melting-point is below the ordinary temperature. 
 RbNO 3 .2HNO 8 decomposes rapidly, and this also must be 
 prepared for analysis in a cool place. The salt RbNO 8 .HNO 8 , 
 and the two caesium salts, are more stable than the others; 
 the monoacid salts are almost stable in the air, and may be 
 preserved indefinitely in hermetically sealed tubes. 
 
ON THE ACID NITRATES. 149 
 
 The melting-points of the acid nitrates, as given by Ditte 
 and found by ourselves, are as follows : 
 
 NH 4 N0 8 .HN0 8 ...... 9 (Ditte) 
 
 KbN0 8 .HN0 8 ...... 62 
 
 CsN0 8 .HN0 8 ...... 100 
 
 KNO S .2HN0 8 ...... -3 (Ditte) 
 
 T1N0 8 .2HN0 8 ...... Undetermined 
 
 NH 4 N0 8 .2HN0 8 ..... 18 (Ditte) 
 
 KbN0 3 .2HN0 3 ...... 39-46 
 
 CsN0 3 .2HN0 8 ...... 32-36 
 
 The melting-points of the rubidium and caesium monoacid 
 nitrates are sharp, but the corresponding diacid salts melt 
 gradually, evidently on account of decomposition. The salt 
 T1NO 3 .2HNO 8 melts below the ordinary temperature, but its 
 exact melting-point was not determined. 
 
 Analyses. The acid nitrates are very easily analyzed. 
 The amount of normal nitrate was determined by simple heat>- 
 ing, and the amount of nitric acid by titration. The follow- 
 ing analyses were made with separate crops of the different 
 compounds : 
 
 Monoacid Rubidium Nitrate, HNO&.RbNO*. 
 
 Calculated. Found. 
 
 i. ii. m. 
 
 HKO S . . . 30.00 30.51 ..." 30.18 
 
 . . . 70.00 . . . 70.43 . . . 
 
 Monoacid Caesium Nitrate, ffNOs- CsNO*. 
 
 Calculated. 
 
 HN0 8 . . . 24.42 24.23 24.13 . . . 24.28 
 
 CsN0 8 . . . 75.58 ...... 76.48 75.52 
 
 Diacid Rubidium Nitrate, 0HNO s .RbNO s . 
 
 Calculated. Found. 
 
 i. n. m. 
 
 HKO S . . . 46.15 45.79 ...... 
 
 KbN0 8 . . . 53.85 . . . 53.74 53.44 
 
 * Determinations by H. P. Beardsley. 
 
150 ON THE ACID NITRATES. 
 
 Diacid Ccesium Nitrate, 
 
 Calculated. Found. 
 
 i. ii. ra. 
 
 HN0 8 .... 39.25 39.23 39.86 . . . 
 CsN0 8 .... 60.75 61.22 
 
 Diacid Thallous Nitrate, TlNOs.2HNO z . 
 
 Calculated. Found. 
 
 i. H. m. 
 
 HN0 8 32.14 33.13 33.02 . . . 
 
 T1N0 8 .... 67.86 67.00 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1901. 
 
INVESTIGATIONS ON DOUBLE NITRATES. 
 I. CAESIUM DOUBLE NITKATES. 
 
 BY H. L. WELLS AND A. P. BEARDSLEY. 
 
 COMPARATIVELY few double nitrates have been described, 
 and it is a curious circumstance that most of these are com- 
 pounds of metals of the rare-earth group, as follows : 
 
 2KN0 8 .Ce(N0 3 ) 8 .2H 2 
 2NH 4 N0 8 .Ce(N0 3 ) 8 .4H 2 
 2NH 4 N0 8 .La(NO 3 ) 8 .4H 2 O 
 2NH 4 N0 8 .Di(N0 8 ) 3 .4H 2 
 
 3Mg(N0 8 ) 2 .2Ce(N0 8 ) 3 .24H 2 O 
 
 3Zn(ko 8 ) 2 .2Ce(N0 8 ) 8 .24H 2 O 
 
 3Mn(N0 3 ) 2 .2Ce(N0 8 ) 3 .24H 2 
 
 3Co(N0 3 ) 2 .2Ce(N0 8 ) 8 .24H 2 
 
 3Ni(N0 8 ) 2 .2Ce (N0 8 ) 8 .24H 2 
 
 3Mg(N0 3 ) 2 .2La(N0 8 ) 8 .24H 2 
 
 3Zn(N0 8 ) 2 .2La(N0 8 ) 3 .24H 2 
 
 3Mn(N0 3 ) 2 .2La(N0 8 ) 8 .24H 2 O 
 
 3Ni(N0 8 ) 2 .2La(N0 8 ) 8 .36H 2 
 
 3Zn(N0 8 ) 2 .2Di(N0 8 ) 8 .60H 2 
 
 3Ni(N0 8 ) 2 .2Di(N0 3 ) 8 .36H 2 O 
 
 3Co(N0 8 ) 2 .2Di(N0 8 ) 8 .48H 2 
 
 2KN0 8 .Ce(NO,) 4 .l^H 2 
 2NH 4 N0 8 .Ce (N0 8 ) 4 .lH 2 
 
 In addition to the above, the only normal double nitrates 
 that have been found mentioned in the literature are the 
 following : * 
 
 * There are known, however, certain easily fusible mixtures of nitrates, 
 such as potassium and lead nitrates, and thallous and silver nitrates, but 
 apparently these do not form definite, crystallized compounds. 
 
152 INVESTIGATIONS ON DOUBLE NITRATES. 
 
 KN0 8 .Au(N0 3 ) 3 
 2KN0 8 .T1(N0 8 ) 8 .H 2 
 
 NH 4 N0 8 .AgN0 8 
 KN0 8 .AgN0 8 
 3KN0 8 .AgN0 8 (?) 
 
 With the exception of the auric and thallic salts, these are 
 all compounds of an alkaline nitrate with a univalent nitrate. 
 It appears to be a singular fact that no double nitrates of 
 alkaline and bivalent metals are known to exist. 
 
 Since caesium is known to form double salts with great 
 facility, we have attempted to prepare double nitrates of this 
 metal with several bivalent metals which generally form double 
 salts very readily. For this purpose we selected lead nitrate, 
 cobalt nitrate, and mercuric nitrate, but after the most careful 
 work we were unable in either case to obtain any definite crys- 
 tallized product, except crops of the simple nitrates : In each 
 case there was evidence of combination from the fact that 
 solutions were obtained which were much too concentrated to 
 hold in solution the caesium nitrate that was present, if it had 
 been uncombined. We conclude, therefore, that double nitrates 
 were formed in solution in these cases, but that they were so 
 exceedingly soluble that they could not be crystallized. 
 
 It seemed desirable to find if trivalent nitrates, other than 
 those of the rare-earth metals, are capable of forming double 
 nitrates with caesium nitrate. For this purpose we selected 
 ferric nitrate, and succeeded, although with some difficulty, in 
 preparing a double nitrate. 
 
 Ccesium Ferric Nitrate, CsNO s .Fe(NO B ) s .7H 2 0.This salt 
 is formed at a rather low temperature in very concentrated 
 solutions containing nitric acid and nearly equal molecular 
 proportions of the component salts. It forms pale yellow, 
 deliquescent, prismatic crystals, which melt, not sharply, at 
 33 -36. The following analyses were made with separate 
 crops : 
 
INVESTIGATIONS ON DOUBLE NITRATES. 153 
 
 Calculated for Found. 
 
 CsN0 3 .Fe(N0 8 ),.7H t O. I. II. HI. 
 
 Caesium nitrate . . 34.64 . . . 35.51 34.87 
 
 Ferric nitrate . . 42.98 42.29 42.73 42.68 
 
 Water 22.38 22.92 
 
 Nitrogen .... 9.94 9.54 
 
 II. CESIUM BISMUTH NITRATE, 2CsNO 3 .Bi(NO 8 ) 8 . 
 
 By G. S. JAMIESON. 
 
 As a continuation of the investigation just described, experi- 
 ments have been conducted with caesium nitrate and bismuth 
 nitrate. The two salts were dissolved in widely varying pro- 
 portions in dilute nitric acid, and the solutions were evapo- 
 rated until crystallization took place on cooling. When 2^ 
 or more molecules of caesium nitrate to 1 molecule of bis- 
 muth nitrate were present, the simple caesium salt crystal- 
 lized out; on the other hand, when the ratio of caesium to 
 bismuth was less than about l to 1, bismuth nitrate was 
 obtained. Between these limits a double salt was produced, 
 particularly upon agitating the cold solution, in the form of 
 colorless, prismatic crystals which are stable upon exposure. 
 These were sometimes more than a centimeter in length. 
 The salt melts at 102. Two crops gave the following 
 analyses : 
 
 Calculated for Found. 
 
 2CsNO 3 .Bi(NO 8 ) 8 . I. II. 
 
 Cgesium nitrate . . . 49.75 47.00 51.22 
 
 Bismuth nitrate . . . 50.25 52.63 48.16 
 
 The analyses do not agree very satisfactorily with each other 
 or with the calculated numbers. This was doubtless due to 
 the syrupy nature of the mother-liquor which made the drying 
 of the crystals by means of filter paper a difficult matter. The 
 appearance of the crystals, in being homogeneous and different 
 from the separate nitrates, however, leaves no doubt that this 
 is a definite double nitrate. 
 
154 INVESTIGATIONS ON DOUBLE NITRATES. 
 
 III. THAU.OUS THALLIC NITRATE, 2TlNO 8 .Tl(NO 8 ) a . 
 
 By F. J. METZGER. 
 
 When thallous nitrate is dissolved in concentrated nitric acid 
 (sp. gr. 1.50) by the aid of heat, a part of the salt is oxidized 
 to thallic nitrate, and, upon cooling the properly concentrated 
 solution, large, colorless, transparent, prismatic crystals are 
 formed. The salt is very stable in dry air, but blackens when 
 exposed to moisture. It melts at 150. Several crops, made 
 under different conditions, were analyzed. 
 
 Calculated for Found. 
 
 2T1N0 3 .T1(NO,) S . I. II. IH. 
 
 Thallic nitrate . . 42.30 43.12 43.13 . . . 
 
 Total thallium . . 66.37 65.30 65.53 65.30 
 
 The thallic nitrate was determined by titrating the iodine 
 set free by the salt from potassium iodide solution. The total 
 thallium was weighed as iodide. 
 
 Conclusions. No evidence has been obtained that alkaline 
 nitrates form crystallizable double nitrates with the nitrates of 
 bivalent metals. 
 
 Several trivalent nitrates, other than those of the rare earth 
 metals, are capable of forming double nitrates. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1901. 
 
ON CAESIUM PERIODATE AND 
 IODATE-PERIODATE. 
 
 BY H.L. WELLS. 
 
 THE two salts to be described were obtained as the result of 
 an attempt to make a thorough investigation of the caesium 
 salts of periodic acid. In view of the well-known complexity 
 of periodates in general, it was expected that a considerable 
 number of caesium periodates would be found, and it was 
 hoped that they might be of much theoretical interest. As 
 a result, however, only normal caesium periodate, CsIO 4 , was 
 obtained when caesium carbonate was added in widely varying 
 proportions to solutions of periodic acid. This was so disap- 
 pointing that experiments upon the addition of periodic acid 
 to caesium hydroxide solutions, which were originally planned, 
 were not carried out. 
 
 Preparation of Periodic Acid. The principles involved hi 
 the method used for this purpose are not new, but as some of 
 the details may be of use to others, a brief description of the 
 process is given. 12.7 g. of iodine are put into a 10 per cent 
 solution of 60 g. of sodium hydroxide. The liquid is heated 
 to boiling in a flask, and a rapid stream of chlorine gas is 
 passed into the continually boiling solution until the large 
 amount of precipitate suddenly formed begins to cause bump- 
 ing, when the flame is instantly removed, and the stream of 
 chlorine is continued until no further increase is observed in 
 the white precipitate of H 3 Na 2 IO<j. While the liquid is still 
 warm the precipitate is collected on a Buchner funnel and then 
 thoroughly washed with cold water. It is dried in a steam- 
 bath. The yield is usually about 22 g. or 80 per cent of the 
 theoretical. The sodium periodate is suspended in a large 
 volume of water, three molecules of silver nitrate in solution 
 
156 ON CAESIUM PERIOD ATE 
 
 are added, the liquid is boiled, filtered hot, and the black pre- 
 cipitate of Ag 8 IO 6 is washed thoroughly with water.* The 
 black silver periodate while still moist is suspended in a 
 small amount of water, and chlorine gas f is passed in persist- 
 ently, with agitation, until the precipitate has become nearly 
 white. The silver chloride is removed by nitration, the 
 liquid is concentrated on the water-bath, then brought to 
 crystallization over sulphuric acid. After being drained, the 
 beautiful crystals of periodic acid are ready for use. 
 
 Ccesium Periodate, CsIO^. This salt crystallizes in white 
 plates, and is rather sparingly soluble in cold water. It was 
 formed upon adding small quantities of caesium carbonate to 
 concentrated solutions of periodic acid, and also when larger 
 quantities, up to an excess of csesium carbonate, were used. 
 It can be recrystallized readily by dissolving in hot water and 
 cooling. When a very large excess of caesium carbonate was 
 added to a solution of the salt and the mixture was evaporated 
 to the point of crystallization, more or less caesium iodate, 
 CsIO 8 , usually was formed on account of an accidental or 
 perhaps spontaneous reduction, but no evidence of the exist- 
 ence of any other periodate was thus obtained. The follow- 
 ing analyses of six separate crops were made, most of them 
 for the purpose of identification : 
 
 
 Calculated fo 
 
 r 
 
 
 JSOl 
 
 ma. 
 
 
 
 Csesium . 
 
 CsIO 4 . 
 
 . 41.05 
 
 I. 
 
 42.02 
 
 ii. 
 40.37 
 
 III. 
 
 IV. 
 
 V. 
 
 VI. 
 
 Iodine 
 
 . 39.20 
 
 
 38.19 
 
 
 
 
 
 Oxygen . 
 
 . 19.75 
 
 19.62 
 
 19.77 
 
 19.61 
 
 19.45 
 
 19.65 
 
 19.20 
 
 Caesium was determined by treating the substance with con- 
 centrated sulphuric acid, evaporating, converting into normal 
 
 * The filtrate contains a considerable amount of periodate which is kept in 
 solution by the nitric acid set free by the reaction. Most of this may be 
 recovered in the form of the golden yellow salt H 3 Ag 2 IO 6 by evaporating the 
 liquid to very small volume and cooling, or after evaporating to dryness on 
 the water-bath another crop of Ag 3 IO 5 may be obtained by treatment with 
 water. 
 
 t According to my experience chlorine is much better than bromine for 
 this purpose. 
 
AND IODATE-PERIODATE. 157 
 
 sulphate by ignition in air containing ammonia, and weighing. 
 Iodine was determined as silver iodide after reduction with 
 sulphurous acid solution. Oxygen was determined by titrat- 
 ing with sodium thiosulphate the iodine set free by the sub- 
 stance in a solution of potassium iodide in the presence of an 
 acid. 
 
 Acid Ccesium lodate-Periodate, HCsIO z IO^H 2 0. This 
 curious salt was first noticed as having crystallized after spon- 
 taneous evaporation from a solution of caesium periodate in 
 dilute periodic acid, in which the periodate had suffered par- 
 tial reduction. A second crop was purposely prepared from 
 a similar solution, and a third product was made by dissolv- 
 ing caesium iodate and periodate in dilute periodic acid and 
 cooling. It forms beautiful, slender white prisms. The 
 three crops gave the following results upon analysis : 
 
 Calculated for Vmu& 
 
 HCBI 2 7 .2H 2 0. "- n 
 
 Hydrogen .... 0.19 
 
 Caesium 24.80 . . . 24.00 . . . 
 
 Iodine 47.39 . . . 47.38 47.11 
 
 Oxygen 20.90 20.75 20.58 . . . 
 
 Water 6.72 
 
 100.00 
 
 Iodine as I 2 5 . . 23.69 23.20 
 
 Iodine as I 2 7 . . 23.69 23.91 
 
 lodic acid was determined by dissolving the substance in 
 dilute nitric acid, precipitating silver iodate by the addition of 
 silver nitrate, and weighing the silver iodide resulting from its 
 ignition. The periodic acid left in the filtrate from the silver 
 iodate was reduced by the addition of sulphurous acid, and the 
 iodine was determined as silver iodide. 
 
ON THE PERIODIC SYSTEM AND THE PROPER- 
 TIES OF INORGANIC COMPOUNDS.* 
 
 BY JAMES LOCKE. 
 
 I. THE INSUFFICIENCY OF THE SYSTEM AS APPLIED TO 
 COMPOUNDS. 
 
 THE problems set before the inorganic chemist by the periodic 
 system, which have directly or indirectly guided almost all 
 the inorganic investigations of the past thirty years, may be 
 divided, roughly in the abstract but very sharply in the results 
 obtained, into two classes. The one, of the fruitfulness of 
 which no one can complain, deals with the superficial group- 
 relations which exist between the members of the various 
 families. Their solution leads to what may be called the 
 development of group analogies. The problems of the second 
 class, on the other hand, have either been entirely neglected, 
 or their study has yielded only the most unsatisfactory results. 
 These have had to do with the contradictions of fact which 
 the system, as formulated by Mendel^eff, involved, and to 
 which the system should in course of time adapt itself. But 
 all the more important inconsistencies which it presented in 
 the year 1870 still remain, and are finally recognized as facts 
 which, for the sake of a convenient principle, must not be 
 forced into a conspicuous position. The behavior of the 
 elements of low atomic weight, the appearance of one and 
 the same element in more than one degree of oxidation, 
 the properties of the platinum metals, either receive no ex- 
 pression in the system, or they stand in direct contradiction to 
 its laws. The difficulties involved, however, are confined 
 chiefly to the compounds. The relations between the elements, 
 as elements, are in almost every case clearly pointed out in the 
 
 * Amer. Chem. Jour., xx, July, 1898. 
 
THE PERIODIC SYSTEM. 159 
 
 usual tables. This is true in regard to both the physical and 
 chemical properties of the elements themselves. In this paper, 
 therefore, only that part of Mendele*efFs law will be discussed, 
 which says that the nature of the compounds of an element is 
 also a function of its atomic weight. 
 
 Of the sixty-five elements having places in the system, nine 
 of them, iron, cobalt, nickel, and the platinum metals, are put 
 in a vertical row by themselves, where they form the so-called 
 eighth family, a family which may be said to fall outside the 
 system proper. In this group many of the regularities which 
 can be traced through other families are wanting. 
 
 The valency of the elements increases by one in each suc- 
 cessive vertical row of the system, until the fourth is reached, 
 after which each family yields two series of compounds, the 
 one of increased, the other of decreased valency. In the 
 potassium-nickel series, for instance, the various elements 
 should enter into compounds with the valencies: 
 
 K 1 , Ca n , Sc m , Ti IV , V v , Cr VI , Mn vn , (F Co, Ki) vm . 
 V m , Or 11 , Mn 1 . 
 
 So, since chromium falls within the sixth family, the great- 
 est weight is laid upon the fact that chromic acid and its salts 
 show such close analogy to sulphuric acid and the sulphates. 
 The fact that potassium permanganate is isomorphous with 
 the perchlorate, and like the latter only sparingly soluble in 
 cold water, likewise assumes the utmost importance. If com- 
 pounds of iron, cobalt, and nickel were known, in which these 
 elements were octavalent, they, too, would possess the greatest 
 theoretical value. But, with the exception of nickel carbonyl, 
 a doubtful case of octavalency, these theoretically necessary 
 compounds have never been obtained. Up to the present no 
 one has succeeded in getting beyond the ferric acid of hexa- 
 valent iron. 
 
 That the decrease in valency, on the other hand, must stop 
 before the eighth family is reached, is obvious, for the seventh 
 family is itself univalent, and a further decrease would prevent 
 iron, cobalt, and nickel from forming any compounds at all. 
 
160 THE PERIODIC SYSTEM 
 
 But this regularity ceases still further on, with chromium. 
 No univalent manganese compounds have been obtained. 
 Now, inasmuch as the iron, cobalt, and nickel compounds fail 
 to conform with a regularity which can otherwise be traced 
 throughout the entire system, they are assigned an entirely 
 different role from that of other elements. They form transi- 
 tion stages, so to speak, from the manganous salts to the 
 copper compounds of the next horizontal row. But this 
 transition is not from the compounds of univalent manganese 
 to those of univalent copper. Univalent manganese is un- 
 known. It is from divalent manganese to divalent copper 
 to copper in a degree of oxidation which in no way corre- 
 sponds to the position of this element in the system, and the 
 salts of which have nothing more in common with the cuprous 
 compounds than has sulphuric acid with hydrogen sulphide ; 
 the same element can be obtained from both. And even then 
 the road is by no means a smooth one, for directly before 
 manganese stands chromium, and since divalent manganese 
 has no place in the system, the chromous salts must be re- 
 garded as the true starting-point for the transition. But the 
 transition from the unstable divalent and stable trivalent de- 
 rivatives of chromium to those of ferrous and ferric iron is 
 through a manganese in which the relative stability in the two 
 degrees of oxidation is exactly the reverse. 
 
 I know I lay myself open here to the charge of expecting 
 too much from the system. But why ? Here is a great law 
 of nature, the correctness of which no one can doubt. But in 
 order to illustrate this law, an empirical arrangement of the 
 elements is formed, and this arrangement we are all too apt to 
 confuse with the law itself. The periodic system is satis- 
 factory enough for the elements. Lothar Meyer's curves of 
 the atomic volumes and melting-points prove this. For the 
 compounds it is unsatisfactory. And in order to find out how 
 the law does apply to compounds, we must sift out and dis- 
 cuss the discrepancies which the system shows in regard to 
 them. 
 
 As with iron, cobalt, and nickel, so it is with all the other 
 
AND INORGANIC COMPOUNDS. 161 
 
 members of the eighth group. These metals, at the time of 
 the promulgation of the periodic system, were assigned a posi- 
 tion which, satisfactory though it was in illustration of their 
 properties as elements, was in regard to their compounds abso- 
 lutely abnormal ; and no amount of investigation and specula- 
 tion, with the periodic system as a basis, can clear up the 
 contradictions which their position entails. 
 
 A second point upon which no light has been thrown lies 
 in the behavior of the so-called " typical elements." These, 
 as Ostwald remarks, instead of being types of the families at 
 the head of which they stand, have in the majority of cases 
 properties directly at variance with those of the other mem- 
 bers of their respective groups. In his " Principles of Chem- 
 istry " Mendele'eff says : " The elements of the first two series 
 have the least atomic weights, and in consequence of this very 
 circumstance, although they bear the general properties of a 
 group, still they show many peculiar and independent proper- 
 ties. These lightest elements are : 
 
 H; 
 
 Li, Be, B, C, N, 0, P- 
 
 Na, Mg, ..,..,. .." 
 
 As Mendele'efT left these elements, " typical," so they have 
 remained, and the history of thirty years contains no mention 
 of a successful attempt to unite them more closely than by 
 that ill-fitting word to the groups to which they should show 
 analogies. 
 
 Lithium is the lightest of all the elements having places in 
 the system. It should, therefore, from the above, have the 
 greatest number of independent properties of any of the 
 typical elements. Or, one should at least expect to find me- 
 tallic lithium and lithium compounds varying from the higher 
 alkali metals and their derivatives in one and the same degree, 
 since their departure from the rest of the group is conditioned 
 by one and the same cause. The abnormal behavior of the 
 other typical elements should be along the same lines as in the 
 case of lithium and its compounds. The essential properties 
 
 11 
 
162 THE PERIODIC SYSTEM 
 
 of metallic lithium stand in strictest analogy to those of the 
 other alkali metals. Metallic beryllium fits in accurately in 
 the series formed by magnesium, zinc, cadmium, and mercury. 
 Between boron and aluminium the analogy is less clear ; but 
 it appears again in full force in the properties of free carbon 
 and silicon. The two elements show, in their physical proper- 
 ties, allotropic modifications, and indifference to reagents, the 
 strongest similarity. Fluorine and chlorine, in the elementary 
 state, present the most striking analogies ; and the gradations 
 in character which the free halogens undergo with increasing 
 atomic weight, from fluorine through to iodine, yield one of 
 the most perfect chemical series imaginable. 
 
 Of the seven elements making up the first horizontal row, 
 therefore, at least four possess, in the elementary state, proper- 
 ties closely akin to those of the subsequent elements in their 
 respective families, and complete the series which the latter 
 form. 
 
 Now, if the periodic system is true for compounds as well 
 as for the elements themselves, then the compounds of anal- 
 ogous elements must be analogous, a simple deduction from 
 the law which is not altered in its bearing by the fact that 
 the elements with which we happen to be dealing fall within 
 the so-called typical class. If the low atomic weights, there- 
 fore, of the four elements lithium, beryllium, carbon, and 
 fluorine are not of such influence as to prevent their ranking 
 as they should in their respective families, we must expect 
 that their compounds will also show analogy to those of the 
 other members of their groups. The compounds of lithium 
 vary more or less from those of the other alkali metals, it is 
 true ; its phosphate and carbonate, especially, are less soluble ; 
 but, nevertheless, the lithium salts in general possess the char- 
 acteristics of alkali derivatives, to an extent which would cer- 
 tainly prevent any one from considering lithium as anything 
 but an alkali metal. Carbon and silicon likewise bear close 
 relationship in their compounds. The property of carbon 
 which gives rise to organic radicals and the existence of such 
 a multitude of combinations between but a few elements, the 
 
AND INORGANIC COMPOUNDS. 163 
 
 ability of the carbon atoms to unite with one another, likewise 
 appears in silicon. The silicon hydrides, the various chlo- 
 rides, bromides, and iodides, silicon-chloroform, and the com- 
 plex oxygen compounds, such as silico-oxalic and mesoxalic 
 acids, illustrate fully the strong similarity between the 
 influence of tetravalent carbon and that of silicon upon their 
 compounds. 
 
 The law, and its expression by the periodic system, there- 
 fore, holds good in the case of lithium and carbon, and no 
 fault can be found with the typical behavior of the two ele- 
 ments. Now, if the effect of small atomic weight upon both 
 lithium and carbon is too slight to cause in the nature of 
 their compounds any great variation from that of their homo- 
 logues, one has a perfect right to assume that this separating 
 influence will not be greater in an element which in its 
 atomic weight lies between the two ; but, after a comparison 
 unprejudiced by the fact that an analogy between the com- 
 pounds of beryllium and magnesium is required by their posi- 
 tions in the system, can a single trace of true resemblance be 
 found between them, or between beryllium derivatives and 
 those of any other element in the second vertical row? Mag- 
 nesium hydroxide and beryllium hydroxide are both dibasic, it 
 is true. But there the similarity ceases. 
 
 A strongly distinctive feature of all magnesium, zinc, and 
 cadmium salts is seen in their behavior toward ammonia and 
 ammonium salts. They also have a pronounced tendency to 
 form double salts with the compounds of other metals. Their 
 oxides are insoluble in water, but soluble even in weak acids. 
 Calcium, strontium, and barium, on the other hand, are char- 
 acterized by their soluble, strongly alkaline hydroxides, by 
 their difficultly soluble sulphates, and by the difficulty with 
 which they enter into double salts. These are group charac- 
 teristics, the presence of any one of which indicates the 
 others. They give a tone to, and define the nature of, all the 
 compounds of the two series. But to beryllium salts not one 
 of them applies. Similar to magnesium though the metal be, 
 its compounds show perfect indifference to ammonium salts ; 
 
164 THE PERIODIC SYSTEM 
 
 they form no double salts,* and the oxide is practically insol- 
 uble even in strong, hot acids; as little soluble, in fact, as 
 alumina. This last fact, together with the ready solubility of 
 the sulphate and of the fluoride, and the insolubility of the 
 sulphide, etc., prevent any comparison of the beryllium com- 
 pounds with those of the alkaline earths. 
 
 With fluorine, the nature of the compounds of which it will 
 be unnecessary to discuss in full, the case is similar, although 
 its atomic weight is nearly three times as great as that of 
 lithium. The properties of the free element show it to be 
 a pronounced halogen, with characteristics which one would 
 expect to find in a halogen of lower atomic weight than chlo- 
 rine. But its compounds are not only widely different from 
 those of chlorine, but they often vary in a direction contrary 
 to that of the gradations observed in the rest of the series. 
 Fluorine, it is true, forms in common with the other halogens 
 an acid of the type HR. But in distinction from hydro- 
 chloric, hydrobromic, and hydriodic acid, hydrofluoric acid has 
 a remarkably low molecular conductivity ; and its salts, in the 
 majority of cases, have properties which are directly the oppo- 
 site of those of the chlorides, bromides, and iodides. The 
 most pronounced chemical characteristic of the halogen acids 
 certainly lies in the solubility of their salts with heavy metals, 
 and the insolubility of the silver, cuprous, and mercurous com- 
 pounds. The fluorides of silver and mercury, on the other 
 hand, are readily soluble : those of calcium, strontium, barium, 
 magnesium, manganese, iron, cobalt, nickel, chromium, cad- 
 mium, copper, bismuth, lead, and others, are either sparingly 
 soluble, or dissolve only in an excess of hydrofluoric acid. In 
 the readiness with which fluorine enters into metallofluoric 
 acids is seen another great distinction from the other halogens. 
 In this, as in many of its other properties, it approaches very 
 closely to cyanogen. Hydrofluoric and hydrocyanic acids and 
 
 * Certain double fluorides must be noted as exceptions to this statement. 
 The double sulphates, phosphates, etc., known, may be disregarded, as they 
 can be represented by simple constitutional formulas, and this formation is 
 of less significance than that of double halogenides and the like. Both 
 calcium and strontium yield double sulphates. 
 
AND INORGANIC COMPOUNDS. 165 
 
 their salts have, in fact, little more in common with hydro- 
 chloric, hydrobromic, hydriodic acids, and the halides, than 
 the fact that they are monobasic or derivatives of monobasic 
 acids. 
 
 We have, therefore, in lithium, beryllium, carbon, and 
 fluorine, four typical elements which in the free state, as ele- 
 ments, stand in close relationship to the other elements of 
 their respective groups. In two of them, lithium and carbon, 
 the influence of their low atomic weights is not such as to 
 deprive their compounds of the chemical nature which charac- 
 terizes the derivatives of their analogues. The others, al- 
 though one of them lies between lithium and carbon, and the 
 other has an atomic weight almost large enough to take it 
 out of the typical class, fail to show analogy to their homo- 
 logues ; that is, to satisfy the requirements of the system, as 
 soon as they pass from the free state into combination. 
 
 These are contradictions which even the mysterious influence 
 of a low atomic weight will not explain. They were known 
 in 1870, and they stand to-day. With the development of 
 organic chemistry, and the increase in the number of bodies 
 known, which could not be explained by the laws of valency 
 alone, stereochemistry appeared. Molecular compounds have 
 been the subject of numerous theories, of which one at least, 
 that of Werner, promises to bear fruit ; but the inconsis- 
 tencies shown, in reference to the other elements, by those of 
 low atomic weight, the exceptions to a law more far-reaching 
 than that of valency itself, have remained untouched, simply 
 in recognition of the fact that it was hopeless to look for any 
 explanation for them which could be reconciled to the system. 
 And why ? Because the inorganic chemist, in standing upon 
 the periodic system, has been unable to dissociate in his mind 
 the chemical compound from the elements of which its formula 
 contains the symbols. 
 
 Of the work done in the development of group analogies, 
 little need be said. This has been the only successful field of 
 investigation which the formulation of the system opened. 
 But it may be well to point out briefly certain inconsistencies 
 
166 THE PERIODIC SYSTEM 
 
 which are involved in the development of these analogies 
 between neighboring metals in the horizontal rows. 
 
 One of the pronounced features of the periodic system is 
 the regularity with which the valency of the elements increases 
 in the successive vertical rows, a regularity to which, in fact, 
 so much importance is attached that Mendele*eff can say with- 
 out fear of criticism, " PbO 2 is the normal salt-forming oxide 
 of lead, as are Bi 2 O 6 , CeO 2 , TeO 8 of bismuth, cerium, and 
 tellurium I " According to Mendele'eff, when an element 
 forms two series of compounds, as copper, for instance, in one 
 of which it has the same valency as its neighbor in the hori- 
 zontal row, its compounds in this degree of oxidation must be 
 similar to those of its neighbor. This rule is well illustrated 
 in the case of copper, for the cupric compounds do bear a very 
 close resemblance to those of the next element, zinc. But 
 examine the rule in its full extension. The formation of an 
 alum by any trivalent metal is rightly regarded as character- 
 izing its behavior throughout all its compounds in this degree 
 of oxidation, and sharply distinguishes its sesquioxide from 
 those of another great class of elements, the rare-earth metals. 
 Associated with the formation of alums is the behavior of the 
 cyanides toward potassium cyanide. All the sesquioxides * 
 which form alums yield soluble double potassium cyanides, 
 the great majority of which have the formula K 3 M(CN) 6 . 
 
 In the series, 
 
 K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, 
 
 the most typical double cyanide is formed by iron; that of 
 cobalt is also stable, and on the other side we find an analo- 
 gous salt formed by manganese, " owing to its proximity to 
 iron." But chromium, beyond manganese, also forms one ; and 
 from some experiments which I myself have made, I find also 
 that potassium vanadicyanide, K 8 V(CN) 6 , is not only capable 
 of existence, but a well-defined compound. We have here the 
 character which the atomic weight 56 lends to the compounds 
 
 * The only exception is in the case of aluminium, the cyanide of which is 
 decomposed by water. Potassium cyanide precipitates the hydroxide. 
 
AND INORGANIC COMPOUNDS. 167 
 
 of trivalent iron exerting an influence upon that of one ele- 
 ment to the right and three elements to the left. With the 
 alums the case is still more remarkable. These bodies are 
 formed by cobalt, iron, manganese, chromium, vanadium, and 
 titanium ; or by six of the ten elements in the row. And not 
 only that; scandium, the metal next to titanium, not only 
 forms a sesquioxide, but this is its only degree of oxidation. 
 Its compounds, however, have the properties characteristic, 
 not of ferric and aluminium salts, but of the rare earths. 
 And, nevertheless, the close proximity of this element to 
 titanium does not lend to the trivalent compounds of the 
 latter a single one of the properties of the rare earths. We 
 must seek the influence to which the character of the titanous 
 salts is due far over at the edge of the table. 
 
 In the foregoing pages no attempt has been made to criti- 
 cise the analogies which exist, in the various series, between 
 the elements themselves. With this feature of the system 
 little fault can be found. The alkali metals, the alkaline 
 earth metals, Be, Mg, Zn, . ., Ge, Sn, Pb, are thoroughly 
 similar among themselves, and well illustrate the gradation in 
 properties with increasing atomic weight. But with the com- 
 pounds the case is entirely different. The derivatives of 
 typical elements show no regularity even in their abnormal 
 behavior; the compounds of the eighth group have a decidedly 
 anomalous position. The similarity between the derivatives 
 of elements in states of oxidation uncalled for by the system, 
 either cannot be explained at all, as in the case of lead and 
 barium, magnesium and manganese compounds, or we are 
 forced to ascribe to one or the other series a certain vague, 
 transitive nature. The reason for this lies, not in the principle 
 that the chemical nature of an element and its compounds is 
 a function of its atomic weight, but in our failure to recognize 
 the twofold character of this principle. We attempt to apply 
 one and the same expression of the law to both elements and 
 compounds. But one has no right, in the systematization of 
 one class of substances, to impose upon himself restrictions 
 which arise only from the system which he employs for 
 
168 THE PERIODIC SYSTEM 
 
 another class. No arrangement of the elements according to 
 their atomic weights can be made, which expresses the anal- 
 ogies between the elements themselves, and in which, for 
 instance, magnesium and manganese receive analogous posi- 
 tions. And the result is that, in order to emphasize the 
 relation between zinc and magnesium, which such an arrange- 
 ment does exhibit, the adherent of the periodic system de- 
 liberately closes the door to an explanation of the far closer 
 analogy between the compounds of magnesium and manganese, 
 as if the latter were not conditioned by an equally important 
 law of nature. 
 
 A system for the compounds, founded upon the atomic 
 weights of the elements, must also necessarily lead to confusion 
 from another source, for the investigation of the compounds is 
 undertaken solely to characterize the general chemical nature of 
 the element. Thus, little distinction is drawn between reac- 
 tions which do not involve a change in degree of oxidation and 
 those which do. If the two are separated at all, it is only 
 when the resulting compounds are very stable, or else con- 
 form to the position of the element in the periodic system. 
 To cuprous compounds, though they are unstable, great im- 
 portance is attached, because copper belongs to the first family. 
 But, even in so full and scientific a text-book as that of Men- 
 dele* eft, the salts of trivalent manganese are reviewed with lit- 
 tle more than the passing remark that they somewhat resemble 
 ferric compounds, and that the chloride decomposes when its 
 solution is warmed. The latter fact seems to deprive the man- 
 ganic salts of all theoretical interest. 
 
 But in a comparison between the derivatives of two metals 
 in the same degree of oxidation, the question of existence be- 
 comes of more importance than stability, for the latter is often 
 dependent entirely upon external conditions. The fact that 
 bromine is liquid, iodine a solid, at ordinary temperatures, 
 does not prevent the most important analogies being drawn 
 between the physical properties of the halogens in the same 
 state of aggregation, though to obtain bromine and iodine as 
 gases, heat must be employed. So it is, with many compounds. 
 
 OF THE 
 
 UNIVERSITY 
 
 Ov, F 
 
 a^/J/. /trrr. 
 
AND INORGANIC COMPOUNDS. 169 
 
 They may perhaps be obtained only under, for us, extraordi- 
 nary conditions, but their very existence may point out in the 
 constituent elements properties which we should otherwise 
 never expect. As an example of such a case only the newly 
 discovered alums of vanadium need be mentioned. The exist- 
 ence of these bodies shows that in spite of the similarity be- 
 tween phosphates and vanadates, the vanadium atom exerts 
 upon the nature of its trivalent compounds the same influence 
 as does iron upon that of the ferric salts. A vanadic solution 
 absorbs oxygen with great avidity, and is therefore called very 
 unstable, but the presence of oxygen is an arbitrarily imposed 
 condition. If oxygen were a rare element, the trivalent salts 
 of vanadium would probably be regarded as its normal com- 
 pounds, and a vanadate looked upon very much as we look 
 upon ferric acid to-day. 
 
 The chemical behavior of an element in a given degree of 
 oxidation must be characterized along two lines. First, by the 
 study of its compounds as mineralogical specimens, their com- 
 position, physical properties, solubility, volatility, and the like, 
 together with the reactions which they show without involv- 
 ing change in the degree of oxidation. Secondly, by its pas- 
 sage from this degree of oxidation into others. The first is, 
 in my opinion, by far the more important, for the properties of 
 the compounds of an element in two different degrees of oxida- 
 tion differ absolutely from one another; and in the two the 
 element appears in entirely different r61es. It is only after we 
 have ascertained in how far the element can alter its apparent 
 character, in corresponding with entirely different elements in 
 different degrees of oxidation, that we can solve its true char- 
 acter and find out a general law for its behavior in all its 
 compounds; and this can be ascertained only by a careful 
 comparison of its compounds with those of other elements 
 in the same degree of oxidation. 
 
 A classification based upon this principle, of course, presents 
 great difficulties, owing to the very unsatisfactory extent to 
 which the elements in their unstable degrees of oxidation have 
 been investigated. A thorough study of the compounds, how- 
 
170 THE PERIODIC SYSTEM 
 
 ever, has shown that it is capable of giving most interesting 
 results. 
 
 II. GRADATIONS IN THE PROPERTIES OF ALUMS.* 
 
 In my first paper on this subject, f I attempted to show 
 that the well-known difficulties which confront the chemist in 
 classifying the elements and their compounds according to the 
 Periodic System are confined practically to the latter class of 
 substances. That there is a definite connection between the 
 properties of an element and those of its compounds is of 
 course obvious, from the numerous cases in which both fulfil 
 the requirements of the system. But, to take one of the chief 
 difficulties, there are almost no instances in which, when an 
 element can appear in more than one degree of oxidation, the 
 properties of its compounds in each type find accurate inter- 
 pretation by the System. As a case in point may be cited ni- 
 tric acid, the nitrates, and nitrous acid and the nitrites, as 
 compared respectively with the derivatives of the pentoxide 
 and trioxide of phosphorus. So far as their oxygen compounds 
 are concerned, the two elements have in neither the quini- 
 valent nor trivalent state anything in common, except their 
 valence and acid-forming nature. In crystallographic form, 
 solubility, even in formulas, their derivatives are inherently 
 different. Nitrogen pentoxide is a volatile liquid ; phospho- 
 rus pentoxide a non-volatile solid. The former forms only the 
 monobasic acid HNO 8 ; the latter yields preferably polybasic 
 acids ; the normal nitrates are in every case readily soluble in 
 water : the normal phosphates, pyrophosphates, and metaphos- 
 phates are in almost every case insoluble in water. 
 
 With the trioxides the degree of similarity is about the 
 same ; strict analogy is practically absent. But nevertheless 
 such a loose periodic-system worship prevails, that almost 
 every modern text-book on inorganic chemistry contains a 
 stereotyped statement that nitric acid is analogous to phos- 
 phoric acid. In some instances even these bounds have been 
 
 * Amer. Chem. Jour., xxvi, August, 1901. t Ibid., xx, 581. 
 
AND INORGANIC COMPOUNDS. 171 
 
 far overstepped. Probably the most thoroughly systematized 
 smaller work on the subject is that of Professor Ramsay. On 
 pp. 333 ff. of his "System of Inorganic Chemistry" (1891), 
 he classes nitrogen tetroxide with vanadium tetroxide, nitrous 
 acid, and with vanadious hydroxide. The latter two are ap- 
 parently analogous because they have the analogous formulas 
 NO(OH) and VO(OH). If this is enough for chemical anal- 
 ogy, why should not the aluminium compound, AIO(OH), be 
 put in the same class ? Aluminium and vanadium both form 
 alums, and, as he emphasizes of vanadium hydroxide, both 
 unite with the alkalies. I speak of this, not in criticism of 
 Professor Ramsay's book, but in protest against the all-prevail- 
 ing tendency to go a-begging for chemical analogy when the 
 system requires it. 
 
 The term " chemical character of an element " has two very 
 different meanings. In the first place it denotes the behavior 
 of the element itself, in the free state ; in the second, the type 
 of reactions of its compounds. In the latter case we may 
 perhaps use an explanatory prefix, and say, for instance, " the 
 character of ferric iron " or of "ferrous iron." The two terms 
 denote two absolutely different states of matter. When one 
 speaks of ferric iron, the mind involuntarily associates it, not 
 with ferrous iron, nor metallic iron, but with the salts of 
 aluminium and trivalent chromium ; with two elements which 
 in the free state bear no analogy to metallic iron,* and no. rela- 
 tion between which and iron is shown by the System. The 
 simplest and most natural analogues of ferric iron are merely 
 the trivalent ions of aluminium and chromium. On the other 
 hand, " ferrous iron " brings to the mind, in addition to the 
 idea of something readily oxidizable, the thought of magnesium 
 and zinc compounds, derivatives of elements which again 
 have nothing in common with metallic iron, and also no re- 
 semblance to the metals which iron in the trivalent state is 
 analogous. Unless a very strong distinction is made between 
 
 * By this I mean the strict analogy which is exhibited between free ele- 
 ments belonging to the same family and sub-group in the System ; for example, 
 metallic zinc and cadmium, silver and gold, arsenic and antimony, etc. 
 
172 THE PERIODIC SYSTEM 
 
 the different usages of the term " chemical character of iron," 
 therefore, this brings about a state of hopeless confusion. Up 
 to the present time it certainly has led to a marked subordi- 
 nation, in importance, of compounds containing elements 
 in unstable states of valence to those in which they are 
 stable. Practically, this is all very well. But from the 
 theoretical standpoint it imposes a direct obstacle in the way 
 of careful and systematic interpretation of fact, for it makes 
 in such cases the power of oxidation or reduction much more 
 important than the simple properties of the compounds as 
 such. 
 
 The laws of electricity cease to obtain, as soon as that force 
 has been converted into light, or heat, or chemical energy 
 no matter which. But ferric iron and ferrous iron are just as 
 distinct primary forms of matter as these are distinct forms of 
 energy. Like the latter, they may be converted the one into 
 the other, or they may be converted into metallic iron. But 
 whether a ferric salt passes into a ferrous salt or into metallic 
 iron, the form of matter analogous to aluminium in the com- 
 bined state ceases to exist. What it becomes, and how readily 
 it undergoes alteration, are questions common to all the ferric 
 compounds, and should be theoretically regarded as belonging 
 to a set of problems entirely distinct and separate from those 
 involving the individual salts. 
 
 In order to clearly characterize an element in a given degree 
 of oxidation, the first step should be, not to ascertain how 
 stable or unstable its compounds are, but how nearly they 
 approach, in their physical properties, solubility, etc., the cor- 
 responding salts of other elements of the same valence. Such 
 arbitrary conditions as, for instance, the presence of oxygen in 
 working with easily oxidizable substances, should be avoided. 
 And, furthermore, the question of analogy should be decided 
 along strict lines. The rare-earth metals are, more often than 
 not, spoken of as analogous to aluminium. But the two nota- 
 ble properties which characterize the aluminium salts the 
 extremely weak basicity of the hydroxide, with the correspond- 
 ing readiness to form basic salts; and the marked tendency 
 
AND INORGANIC COMPOUNDS. 173 
 
 to form complex ions (still more highly developed in chromium 
 and ferric iron), as illustrated in the behavior of aluminium 
 solutions containing oxalic, tartaric, or other organic acids, are 
 just the reverse of the properties which characterize the 
 strongly basic rare-earths. In Dammer's Handbuch we find 
 specific mention of fourteen derivatives of aluminium chloride. 
 Under cerous chloride the chlorine compound of the best 
 investigated rare earth we find but two.* Barium chloride 
 and cupric chloride show a similar difference. In both cases 
 the divergence is due to much the same cause. The very 
 nature of the strongly basic ions of barium and cerium pre- 
 vent the formation of such derivatives ; and since the diver- 
 gence in the behavior of the chlorides is duplicated in that of 
 sulphates, nitrates, bromides, and countless other salts, it 
 surely indicates a radical distinction in the characters of 
 cerium and aluminium, to which much more weight should be 
 attached than to the fact that the hydroxides of both are 
 flocculent precipitates. The latter resemblance is essentially 
 the basis of assertions that they are analogous. Cerium may 
 be regarded as the prototype of the rare-earth metals ; alumin- 
 ium as that of trivalent chromium, manganese, vanadium, iron, 
 cobalt, and others. In each group the reactions of these typi- 
 cal metals are closely followed by all their respective mem- 
 bers, in a way which proves that the separate reactions of each 
 are not to be regarded as individual functions of the elements, 
 but closely correlated with one another. For instance, it 
 means that a trivalent metal which forms alums will yield 
 complex tartrate ions ; and that a trivalent metal which has a 
 strongly basic hydroxide will form an insoluble oxalate, and 
 not form an aluin nor a double chloride with potassium chlo- 
 ride ; but that its oxide will dissolve readily in dilute acids. 
 As illustrations in the above, I have chosen the trivalent 
 metals by chance. Others would have done just as well. If 
 we compare the solubility relations and, exclusive of those in 
 which a change of valence takes place, the reactions, of cu- 
 prous compounds, with those of the corresponding silver salts, 
 
 * Exclusive of the platinichloride, etc. 
 
174 THE PERIODIC SYSTEM 
 
 we find an almost exact similarity between the two. But to 
 ascertain experimentally in how far the strict classification of 
 the elements in the manner indicated is possible, the trivalent 
 metals are especially suitable. Among the alum-forming ele- 
 ments we find one (cobalt) which in the trivalent state yields 
 compounds that are very unstable because of the high valence ; 
 and a second one (vanadium) whose compounds are equally 
 unstable because of the abnormally low valence. The range 
 of atomic weights of the alum-forming trivalent metals, from 
 27 (Al.) to 114 (In.) is also exceptionally great for an iso- 
 morphous series, and these elements constitute members of 
 six out of the eight families in the System. 
 
 The total number of trivalent elements entering into the 
 compounds M 1 M ra (SO 4 ) 2 .12H 2 O, is nine, aluminium, tita- 
 nium, vanadium, chromium, manganese, iron, cobalt, gal- 
 lium, and indium. Of univalent elements, M 1 represents five : 
 sodium, potassium, rubidium, caesium, thallium ; and also the 
 ammonium, hydroxylammonium, methylammonium, etc., radi- 
 cals.* A cursory glance over the literature shows at once that 
 there is a more or less pronounced decrease in stability with 
 increasing atomic weight of the trivalent metal in an alum, and 
 apparently that the heavier the univalent metal, the greater will 
 be the stability. Thus, sodium enters into alums only with 
 the lightest of the trivalent metals, aluminium, vanadium, and 
 chromium. The only stable potassium alums are those of 
 aluminium and chromium : potassium ferric and gallium alums 
 break down into basic salts very readily, and the only double 
 sulphates of potassium and indium which can be obtained have 
 the formulas Kin (SO 4 ) 2 .4H 2 O and K 2 SO 4 .2In 2 (SO 4 ) 3 . 6H 2 O. 
 
 The only alums which have as yet been obtained of tita- 
 nium are those in which the univalent metal is caesium or 
 rubidium.! After much doubt had been cast upon the exist- 
 
 * The existence of the silver aluminium alum described by Church and 
 Northcote is denied by Retgers. 
 
 t Piccini, Zeitschr. fiir anorg. Chemie, xvii, 355. I have obtained indica- 
 tions of the existence of that of potassium. 
 
AND INORGANIC COMPOUNDS. 175 
 
 ence of the manganese alums described by Mitscherlich, Piccini 
 has shown that at least the csesium compound of this element 
 can be easily obtained.* Owing to the great instability of 
 their salts, and the correspondingly slight value which could 
 be attached to measurements of their physical constants, I 
 have felt it useless to include these metals in the present 
 investigation, which covers, however, alums of all the other 
 trivalent metals mentioned above, except gallium. 
 
 I have also rejected the salts of the substituted amines, as 
 involving the ulterior question of the influence of complex 
 ra/licals. The ammonium alums, however, in view of their 
 frequent occurrence, have been carefully studied, in order to 
 ascertain how far the generally greater readiness of this com- 
 plex to yield double salts, as compared with that of potassium, 
 is capable of exact determination. The univalent groups 
 studied, therefore, were sodium, potassium, ammonium, ru- 
 bidium, csesium, and thallium. For the necessary rubidium 
 and csesium material I am indebted to the kindness of Prof. 
 H. L. Wells, who kindly placed some very pure compounds 
 of these metals at my disposal. The salts given in the follow- 
 ing table were investigated. 
 
 Al=:27. V = 51.1. Cr :r: 52.1. Fe = 66. Co = 59. In 113. 
 
 23 ISTa . . Na 
 
 39 K K K K 
 
 NH 4 NH 4 NH 4 NH 4 
 
 85 Kb Kb Kb Rb Kb Kb 
 
 133 Cs Cs Cs Cs Cs Cs 
 
 204 Tl Tl Tl Tl 
 
 Solubility in Water at 25. As a criterion of the properties 
 of these compounds, their solubility in water at 25 was first 
 determined. The literature contains data only upon the solu- 
 bility of the compounds of aluminium f and vanadium, J and 
 all those given seem to be of more or less doubtful character, 
 
 * Zeitschr. f iir anorg. Chemie, xvii, 355 ; xx, 12. 
 
 t Poggiale, A., ch. (3), viii, 467; Setterberg, Liebig's Annalen, ccxi, 100; 
 Mulder, Scheidekund, Verhandl, 1864, 91 ; etc. 
 t Piccini, Zeitschr. fur anorg. Chemie, xiii, 441. 
 
176 THE PERIODIC SYSTEM 
 
 the results having in each case been obtained without the use 
 of a thermostat. In addition to this fact, the actual temper- 
 atures chosen by different experimenters for their determina- 
 tions were rarely identical; so that their results for a given 
 degree would have to be calculated by interpolation. I have 
 accordingly been forced to make new measurements in each 
 case. 
 
 The determinations were made in two-ounce bottles, con- 
 taining water and the finely divided salts in excess. These 
 were placed in a digesting bath as described by Noyes,* and 
 rapidly shaken by means of a turbine for from four to six 
 hours. The bottles were then placed upright in the bath, 
 and the suspended salts allowed to settle for two hours. The 
 temperature of the bath was controlled by a gas regulator, and 
 its variation, in the determinations used below, never exceeded 
 0.2. From three to eight cubic centimeters of the superna- 
 tant solutions were then transferred by means of a pipette, the 
 bulb of which had been warmed, to small weighing-glasses, 
 weighed, and evaporated to dryness. The amounts of the dis- 
 solved salts were then determined by heating for four hours 
 at 200, at which temperature they became fully dehydrated. 
 The accuracy of this method was first ascertained by experi- 
 ments made with a variety of alums. In these a weighed 
 quantity of the salt was dissolved in a few cubic* centimeters 
 of water, evaporated to dryness, and heated to constant weight. 
 The water was added to ascertain whether an error would be 
 introduced through basic salt formation. The following are 
 some of these determinations : 
 
 0.4513 g. CsCr(S0 4 ) 2 .12H 2 gave 0.2858 g. anhydrous salt, 
 
 calculated 0.2869 g. 
 0.3540 g. CsFe(S0 4 ) 2 .12H 2 gave 0.2262 g. anhydrous salt, 
 
 calculated 0.2259 g. 
 0.2264 g. CsAl(S0 4 ) 2 .12H 2 gave 0.1411 g. anhydrous salt, 
 
 calculated 0.1403 g. 
 0.7814 g. TlCr(S0 4 ) 2 .12H 2 gave 0.5473 g. anhydrous salt, 
 
 calculated 0.5474 g. 
 
 * Zeitschr. fur physikal. Chemie, ix, 606. 
 
AND INORGANIC COMPOUNDS. 177 
 
 The solubility determination of the vanadium alums re- 
 quired great caution, owing to the readiness with which they 
 undergo oxidation. These salts, which were made according 
 to Piccini's directions,* by electrolysis, were washed by decan- 
 tation with boiled water containing carbonic acid, in ttfe vessel 
 in which they had been crystallized ; and were then brought 
 directly into the bottles, without having been exposed to the 
 air for an appreciable length of time. The water used for 
 their solution had previously been boiled, and carbon dioxide 
 led through it in the bottles. The resulting solutions were in 
 this case analyzed by the titration of weighed quantities 
 according to Browning's method, f The alums of cobalt, of 
 which only the potassium and ammonium compounds had been 
 described hitherto, J unfortunately decompose quite rapidly in 
 solution, giving off oxygen, and determinations could therefore 
 not be made with them. My supply of indium, also, was too 
 small to permit a determination of any but its caesium alum. 
 The results obtained with this compound alone, however, suffi- 
 ciently indicate the role of indium as an alum-forming element.- 
 
 The results given below are in about half the cases averages 
 of two or more closely agreeing determinations, and are chosen 
 from the total number made as having been obtained under 
 the most perfect conditions of temperature, etc. The values 
 refer to the number of parts of salt dissolved in one liter of 
 water, and in the successive tables M 1 represents the univa- 
 lent metals contained in the alums of the trivalent metal M m . 
 
 M m = Aluminium, At. wt. = 27. 
 
 M 1 Na K NH 4 Tl Kb Cs 
 
 Anhydrous salt, . . 72.3 91.9 75.0 18.1 4.7 
 
 Hydrated salt, . . 138.4 191.9 117.8 31.5 7.6 
 
 M ra = Vanadium, At. wt. = 51.1. 
 
 M 1 = Na K NH 4 Tl Kb Cs 
 
 Anhydrous salt, . . . . 316.9 256.0 57.9 7.71 
 
 Hydrated salt, . . 785.0 433.1 99.3 13.10 
 
 * Zeitschr. fur anorg. Chemie, xi, 106. 
 
 t Ibid., i, 158. $ Soc. Edinburgh, lix, 760. 
 
 Piccini obtains the same order for the solubility of the vanadium alums 
 
 12 
 
178 THE PERIODIC SYSTEM 
 
 M ra = Chromium, At. wt. = 52.1. 
 
 M 1 = Na K NH 4 Tl Kb Cs 
 
 Anhydrous salt, . . 125.1 107.8 104.8 25.67 5.7 
 
 Hydrated salt, . . 243.9 212.1 163.8 43.40 9.4 
 
 M m = Iron, At. wt. = 56. 
 
 M 1 = Na K NH 4 Tl Eb Cs 
 
 Anhydrous salt, Does not Basic salt 441.5 361.5 97.4 17.1 
 
 exist. separated. 
 
 Hydrated salt, 1244. 646.0 169.8 27.2 
 
 M m = Indium, At. wt. = 114. 
 
 M 1 Na K NH 4 Tl Eb Cs 
 
 Anhydrous salt, Does not Does not . . . . . . 75.7 
 
 exist. exist. 
 
 Hydrated salt, 117.3 
 
 The sodium alums known dissolve in much less than their 
 own weight of water, and were therefore unsuitable for exact 
 determinations. A sodium alum of iron could not be obtained. 
 With regard to the effect of the univalent metals upon the 
 solubility, it is seen that the higher the atomic weight of an 
 alkali-metal, the less soluble is its alum with any given triva- 
 lent metal. The solubilities of the ammonium and thallium 
 compounds, however, lie between those of the corresponding 
 salts of potassium and rubidium. The salt NH 4 A1(SO 4 ) 2 . 
 12H 2 O forms the only exception to this rule. It is some- 
 what more soluble than KA1(SO 4 ) 2 .12H 2 O.* The thallium 
 alums, furthermore, are in each case somewhat less soluble 
 than those of ammonium. 
 
 The key to the effect of the trivalent elements upon the 
 
 at 10. Expressed in parts by weight of the salts dissolved in 100 parts water, 
 his determinations are as follows : 
 
 V K V NH 4 V Tl V Rb V Cs 
 
 198.4 39.76 11.06 2.56 0.464 
 
 * In Poggiale's determinations, this ammonium alum was found to be less 
 soluble than the potassium salt. His results for the latter correspond almost 
 exactly with mine for the ammonium salt, and those for the ammonium alum 
 with mine for that of potassium. It is probable that the coincidence is due 
 to a false arrangement of his tables. 
 
AND INORGANIC COMPOUNDS. 179 
 
 solubility relations is given in the failure of any beyond 
 vanadium to form alums with sodium, and in the decomposi- 
 tion of the potassium ferric salt by water. As this would 
 indicate, the solubilities of the alums of the successive trivalent 
 metals with a given univalent metal increase steadily from 
 aluminium to indium. From what one can judge through the 
 literature, gallium alums, also, follow this rule, being, next to 
 those of indium, the most soluble. The chromium alums 
 alone form an exception. Their behavior will be touched upon 
 at a later point. 
 
 In the following table I give the solubilities of all the 
 alums determined, expressed in gram-molecules of the anhy- 
 drous salts per liter of water: 
 
 M 1 = K 
 
 M 111 = Al 0.28 
 V 
 
 Cr 0.441 
 Fe 
 In 
 
 These results may be represented graphically as a function 
 either of the molecular weight of the respective alums of 
 given trivalent metals, which is virtually the same thing as 
 if they were plotted according to the atomic weights of the 
 univalent metals, but introduces the ammonium salts ; or they 
 may be referred to the atomic weights of the trivalent metals. 
 To show the influence of both the univalent and trivalent 
 metals, I give in Figs. I. and II. the results obtained by each 
 method. In Fig. I. the solubilities are plotted as functions 
 of the molecular weights, the univalent metal being the 
 variable. 
 
 As is seen, the curves are in general of the same nature, 
 but they have no uniformity of position, and render no 
 gradation in the solubilities of the alums visible. This is true 
 whether we include the ammonium salts or not, or, in other 
 words, whether the solubilities be referred to the molecular 
 
 NH 4 
 
 Tl 
 
 Rb 
 
 Cs 
 
 0.387 
 1.210 
 
 0.177 
 0.573 
 
 0.059 
 0.177 
 
 0.013 
 0.0204 
 
 0.407 
 
 0.212 
 
 0.078 
 
 0.0151 
 
 1.659 
 
 0.799 
 
 0.293 
 
 0.045 
 0.172 
 
180 
 
 THE PERIODIC SYSTEM 
 
 
 
 weights or to the atomic weights of the univalent metals. 
 The only influence which the univalent metals or radicals exert 
 is a specific one. The caesium salts are in all cases the least 
 soluble, the rubidium salts next, but the comparatively great 
 solubility of the thallous compounds, which, according to the 
 
AND INORGANIC COMPOUNDS. 181 
 
 gradation observed in the case of the compounds of the alkali 
 metals proper, should be the least soluble of all, prevents any 
 satisfactory representation of the solubilities as a function of 
 the atomic weights of the univalent metals. 
 
 This specific influence of the univalent metals is readily 
 seen in Fig. II, in which the solubilities are plotted as a func- 
 tion of the atomic weights of the trivalent metals. 
 
 As we pass from the caesium compounds through those of 
 rubidium and thallium to the ammonium alums, the difference 
 in solubility of the successive salts of a given trivalent metal 
 becomes greater. The increase is also the greater, the heavier 
 the trivalent metal; the result being that while the caesium 
 alums of all the trivalent metals differ only slightly in solu- 
 bility, and their curve has therefore a nearly horizontal posi- 
 tion, the curves of the remaining univalent radicals successively 
 become more inclined ; * the whole effect being somewhat like 
 that of an opening fan. But in each case the curves show a 
 sharp break at the chromic compounds ; slight in the caesium 
 curve, but more pronounced with each successive univalent 
 metal. Apparently, since the chromic salts accordingly fail 
 to fall properly within the series of gradations with increasing 
 atomic weight of the trivalent metal, the conclusion might be 
 drawn that the great readiness with which this element forms 
 alums is in the actuality an abnormal property. I believe, 
 however, that this is a very loose way of interpreting the 
 fact. Abnormal things do not occur in nature. In the in- 
 creasing solubilities of the chromic alums as we pass from 
 caesium to ammonium, they show exactly the same mutual 
 relations as are shown by the alums of aluminium, vanadium, 
 or iron. They are perfectly normal in everything, except that 
 their behavior is not expressed, like that of the other trivalent 
 metals, by the atomic weight of the metal they contain. And 
 
 * Prof. Horace L. Wells suggests to me that the marked difference 
 shown here between the solubilities of the rubidium and caesium alums of 
 iron might have a practical bearing upon the technical separation of ru- 
 bidium and caesium. At present this is usually effected by the fractional 
 crystallization of their aluminium alums. By using the ferric compounds 
 it would probably be much more perfect. 
 
182 
 
 THE PERIODIC SYSTEM 
 
 30 
 
 'Weights,. 
 
 FIGURE II. 
 
 51 52 
 
 56 
 
 this fact indicates almost positively that the properties of the 
 compounds are not a function of the atomic weight of the element. 
 
AND INORGANIC COMPOUNDS. 183 
 
 If they were, chromic alums would have about twice their 
 actual solubility. 
 
 I do not mean by this to make a sweeping denial of all 
 analogy shown in the Periodic System. Whatever the deter- 
 mining cause of the properties of compounds may be, it is 
 certainly commensurable, in a great many instances, with the 
 atomic weights. But our progress toward the systematization 
 of the compounds will be slow indeed, if we refuse to recog- 
 nize that Mendele'efFs law is but a slight approximation of the 
 truth. The abnormal behavior of the thallium alums, accord- 
 ing to the Periodic System, might be conventionally explained 
 by the position of thallium as extraneous to the first family of 
 the table. But in the case of chromium such an explanation 
 is totally inapplicable. The elements aluminium, vanadium, 
 iron, and cobalt * constitute members of three different fami- 
 lies, and nevertheless their alums show a satisfactory grada- 
 tion according to the atomic weights of these elements. 
 Their alums are merely intermediate members in the series 
 of which the end members, the aluminium, gallium, and indium 
 salts, are so conspicuous because they are compounds of 
 three elements in the same family in the system, and thus 
 exhibit the gradations required by the latter. 
 
 The chromic alums, the thallous alums, and the ammonium 
 alums exhibit in each case the same peculiar relations to one 
 another as do the alums of their kindred elements with normal 
 behavior. In the case of the ammonium alums this is espe- 
 cially important, for it shows that whatever the determining 
 cause of the properties of the alums may be, exactly the same 
 cause must le ascribed to the behavior of the ammonium salts as 
 to that of the alums containing only metals : since the am- 
 monium radical has no true atomic weight, therefore the 
 atomic weight itself is not the determining cause in the case 
 of any alums whatsoever. 
 
 It is an interesting fact that the molecular volumes of the 
 alums, when arranged according to their ascending values, 
 
 * The effect of this element upon its alums is shown by their melting- 
 points, as will be described later. 
 
184 
 
 THE PERIODIC SYSTEM 
 
 bring both the trivalent and univalent elements into the same 
 order as do their solubilities. In the following tables* the 
 molecular volumes are given as calculated from Soret's deter- 
 minations, with the double formulas, M 2 SO 4 , M 2 (SO 4 ) 8 . 24H 2 O 
 as the basis of the molecular weights. The solubilities are 
 expressed in gram-molecules per liter. 
 
 
 M m =Al. M m 
 
 = Cr. 
 
 M ra = Pe. 
 
 
 Mol. vol. Solubility. Mol. vol. 
 
 Solubility. 
 
 Mol. vol. Solubility. 
 
 M 1 ~K 
 
 546.9 0.28 550.8 
 
 0.441 
 
 
 JL\. 
 
 NH 4 
 
 555.9 0.387 557.6 
 
 0.407 
 
 562.8 1.659 
 
 Tl 
 
 566.6 0.177 557.7 
 
 0.234 
 
 560.2 0.799 
 
 Eb 
 
 562.2 0.059 561.7 
 
 0.079 
 
 573.3 0.376 
 
 Cs 
 
 579.2 0.013 581.8 
 
 0.015 
 
 579.3 0.045 
 
 
 M 1 = Cs. 
 
 M' = 
 
 (NH 4 ). 
 
 
 Mol. vol. Solubility. 
 
 Mol. vol. 
 
 Solubility. 
 
 M m = 
 
 Al 578 0.013 
 
 555.9 
 
 0.387 
 
 
 Cr 582 0.015 
 
 557.6 
 
 0.407 
 
 
 Fe 579 0.045 
 
 562.8 
 
 1.659 
 
 
 In 584 0.172 
 
 ... 
 
 . . 
 
 The coincidence, while not absolute, shows that there is a 
 close connection between the molecular volumes of the alums 
 and their solubility. The variations, furthermore, lose much in 
 weight in the fact that a very small error in the determination 
 of the specific gravity is greatly multiplied in the quotient 
 
 molecular weight 
 specific gravity 
 
 Solubility at Different Temperatures. The next question 
 to be decided was whether there is any regularity in the 
 effect of increasing temperature upon the solubility of dif- 
 ferent alums. Here I was much limited in my choice of salts, 
 insomuch as the ferric compounds break down into basic 
 salts when treated with pure water at higher temperatures, 
 
 * Soret, Arch. sc. phys. nat. Geneve, xii, 563, etc. Arzruni, Physikalische 
 Chemie der Krystalle (1893), p. 130. Soret's original article is not at my 
 disposal. 
 
AND INORGANIC COMPOUNDS. 
 
 185 
 
 10 
 
 30 
 
 Temperature 
 
 FIGURE HI. 
 
 40 
 
186 
 
 THE PERIODIC SYSTEM 
 
 and the chromic compounds pass into their green modifi- 
 cations at about 50. This behavior of the ferric alums pre- 
 vented the comparative study of any but the rubidium and 
 caesium compounds. As I wished merely to find the general 
 direction of the solubility curves, I did not carry out deter- 
 minations at temperatures below 25. Experiments on 
 chromic salts showed that even at 45 they became slightly 
 green on prolonged digestion, and 40 was therefore taken as 
 the upper limit. The following table gives the solubilities of 
 the rubidium and caesium alums of aluminium, chromium, and 
 iron, at intervals of 5 between these limits. 
 
 Eb Al. 
 
 Bb Or. 
 
 Eb Pe. 
 
 25 
 
 Pts. per 
 liter. 
 
 18.1 
 
 Gr. mol. 
 per liter. 
 
 0.059 
 
 Pts. per 
 liter. 
 
 25.7 
 
 Or. mol. 
 per liter. 
 
 0.078 
 
 Pts. per 
 
 liter. 
 
 125.4 
 
 Gr. mol. 
 per liter. 
 
 0.294 
 
 30 
 
 21.9 
 
 0.072 
 
 31.7 
 
 0.096 
 
 202.4 
 
 0.617 
 
 35 
 
 26.6 
 
 0.087 
 
 41.1 
 
 0.128 
 
 Basic salt 
 
 
 
 
 
 
 
 separated. 
 
 
 40< 
 
 25 
 
 30 
 35 
 40 
 
 32.2 0.106 59.7 0.181 
 
 Cs-Al. 
 
 4.70 0.0130 
 
 5.89 0.0167 
 
 7.29 0.0207 
 
 9.00 0.0256 
 
 Cs Cr. 
 
 5.70 0.0150 
 
 9.60 0.0250 
 
 12.06 0.0320 
 
 15.30 0.0405 
 
 Cs-Pe. 
 
 17.1 0.045 
 
 25.2 0.066 
 37.5 0.099 
 60.4 0.156 
 
 Fig. Ill shows the curves corresponding to the solubilities 
 of these salts, expressed in gram-molecules per liter. To 
 facilitate comparison the solubilities curves of the rubidium 
 salts are dotted. 
 
 There seems here to be no marked connection between the 
 various alums with respect to the univalent metals ; the curve 
 of CsFe(SO 4 ) 2 crossing that of RbAl(SO 4 ) 2 at about 33. 
 But if we compare the alums first of caesium and then of 
 rubidium, we see that the more soluble the alum of a given 
 univalent metal is at 25, the more rapidly does its solubility 
 increase with the temperature. This is marked even in the 
 case of the caesium salts of aluminium and chromium, the 
 
AND INORGANIC COMPOUNDS. 187 
 
 solubilities of which at 25 are almost the same (0.013 and 
 0.015 gr. mol.) The curve of RbFe(SO 4 ) 2 could be followed 
 only to 30, as at 35 a sparingly soluble basic salt separated 
 from the solution; and the extreme difference in solubility 
 shown for the interval 25-30, may in part be ascribed to a 
 hydrolytic action. 
 
 Melting-points. The melting-point of a hydrated salt is of 
 course intimately connected with its solubility; so that in 
 studying them one would expect results approximately anal- 
 ogous to the above. But at the same time I was enabled hi 
 this way to examine a greater variety of alums, including 
 those of cobalt, the solubility of which could not be deter- 
 mined, the rubidium indium salt, and the potassium-vanadium 
 and potassium-ferric compounds. 
 
 The cobaltic salts examined were those containing rubidium 
 and caesium. These were made in the same way as Marshall * 
 obtained the ammonium and potassium cobaltic alums, by the 
 electrolytic oxidation of cobaltous sulphate and addition of 
 the calculated amount of the salt of the desired alkali metal. 
 Both compounds came down in coarse crystalline grains, which 
 could be readily washed by lixiviation with cold water. They 
 were sufficiently stable to permit of rapid drying in the air, 
 but on standing for any length of time they lost oxygen and 
 yielded cobaltous sulphate. The rubidium alum decomposed 
 in this matter even in a sealed specimen tube, on standing for 
 a few months. The caesium compound, however, when pre- 
 served in the same way, has still retained its green color ; 
 thus forming an interesting example of the greater stability 
 of caesium double salts over those of rubidium. Both com- 
 pounds crystallized in glistening microscopic octahedrons. In 
 view of their marked physical properties, it was not deemed 
 necessary to submit them to a full analysis, but as a matter of 
 general precaution a weighed quantity of each was ignited be- 
 low a red heat, and the amount of the residual cobaltous and 
 alkali-sulphates determined. 
 
 * Soc. Edinburgh, lix, 760. 
 
188 THE PERIODIC SYSTEM 
 
 0.0683 g. RbCo(S0 4 ) 2 .12H 2 gave 0.0361 g. Rb 2 S0 4 + CoS0 4 
 
 calculated 0.0357 g. 
 
 0.7004 g. CsCo(S04) 2 .12H 2 gave 0.2962 g. Cs 2 S0 4 + CoS0 4 
 
 calculated 0.2959 g. 
 
 The determination of the melting-points was carried out in 
 sealed capillary tubes, of various diameters, and about two 
 inches in length. The tubes were in each case filled to some- 
 what over half their height with the powdered salts. At least 
 five determinations were made on each compound, and samples 
 from several different preparations used for each. Variations 
 in the manner of heating, etc., introduce comparatively large 
 errors in the melting-points, which are in no case very sharp. 
 But after some practice, I was able, by carefully following one 
 method of procedure, to reduce the error to about 1 ; and in 
 cases where two alums melted at about the same point, as the 
 NH 4 A1 (95) and NH 4 Cr (94) salts, comparative determina- 
 tions were made side by side in the same apparatus. So while 
 no pretence to great accuracy in these determinations is made, 
 the errors involved are at least about the same for each com- 
 pound, and thus the data given below are not vitiated for the 
 purpose of comparison. 
 
 The literature contains various notices on the melting-points 
 of aluminium- and chromium alums, and Piccini also deter- 
 mined approximately those of the vanadium salts with sodium 
 (9), potassium (20), and ammonium (50). The last I find 
 to melt at 45, and for vanadium potassium alum I have ob- 
 tained data anywhere from 20 to 28. The purification of 
 this compound is so difficult that a constant melting-point 
 cannot be observed. My results on the aluminium alums cor- 
 respond fairly well with those of Erdmann,* though being 
 slightly higher ; and both his and mine are much higher than 
 those determined by Tilden.f 
 
 Tilden . . 
 
 Na Al ; 
 
 . 61 
 
 K Al; 
 
 84.5 
 
 NH 4 Al ; 
 92 
 
 Rb Al ; 
 99 
 
 Cs Al. 
 
 106 
 
 Erdmann . 
 
 . . . 
 
 92.5 
 
 . 
 
 105 
 
 120.5 
 
 Locke . . 
 
 . 63 
 
 91 
 
 95 
 
 109 
 
 122 
 
 * An. Pharm., ccxxxii, 3. f London Soc. Trans., xlv, 266. 
 
125 
 120 
 115 
 
 105 
 
 Al) 
 
 Al 
 
 Al 
 
 75 
 
 05 
 
 55 
 
 45 
 
 35 
 
 Atomic "Weights, 
 
AND INORGANIC COMPOUNDS. 189 
 
 Observations on sodium-chromium and csesium-indium alums 
 gave no satisfactory results ; the fusion process in each case 
 covering a range of about ten degrees, and in the case of the 
 indium salt, being even then imperfect. My complete results 
 follow : 
 
 M m = Al. V. Cr. Fe. Co. In. 
 
 M 1 = Na 63 (9)* 
 
 K 91 (20)* 89 28 
 
 Tl 91 48 92 37 
 
 NH 4 95 45 94 40 . . (36)f 
 
 Rb 109 64 107 53 47 42 
 
 Cs 122 82 116 71 63 
 
 The order in which the trivalent metals fall, according to 
 the melting points of their alums with a given univalent metal, 
 is here the same as when determined by the solubility relations 
 of their salts. The one exception lies in the doubtful melting 
 point of vanadium potassium alum, which is a few degrees 
 lower than that of the corresponding ferric salt. The cobalt 
 alums melt at a slightly lower .temperature than do the iron 
 compounds, and indium rubidium alum lower than any other 
 of the rubidium salts, just as the indium csesium alum is the 
 most soluble member of the caesium series. 
 
 With reference to the univalent metals also, the same grad- 
 uation obtains as in the solubilities, but only in so far as the 
 alkali metals are concerned. The order of the thallium and 
 ammonium series is changed, the thallium salts lying next 
 those of potassium. The vanadium compounds here form the 
 only exception, and in view of their great instability it is to be 
 doubted whether this is not due to unavoidable impurities. 
 
 In Figure IV, I give the melting-points plotted as a func- 
 tion of the atomic weights of the trivalent metals, R m . 
 
 Here again we see a sharp break in the case of the chromic 
 salts. The melting points of the csesium and rubidium alums 
 of aluminium, vanadium, iron, and cobalt are represented by 
 nearly parallel straight lines, which would indicate a very 
 
 * Piccini, 1. c. t J. Prakt. Chemie (2), vii, 14. 
 
190 THE PERIODIC SYSTEM 
 
 simple relation between them, and the atomic weights of these 
 metals. But we must either admit that the values of the 
 melting-points are not dependable upon the atomic weights, 
 or that the chromic alums form an exception to a law of 
 nature. 
 
 III. THE SOLUBILITIES OF ALUMS AS A FUNCTION OF 
 Two VARIABLES.* 
 
 In my second paper in this series f I attempted to show that 
 a definite gradation can be traced in the solubility of the 
 alums of aluminium, vanadium, chromium, and iron, severally, 
 with ammonium, thallium, rubidium, and caesium. When the 
 solubilities at 25 of the sixteen compounds considered, ex- 
 pressed in gram-molecules per liter of water, are plotted as a 
 function of the atomic weights of the trivalent metals, a fig- 
 ure of remarkable regularity is obtained. The solubilities of 
 the alums of aluminium, vanadium, and iron, with any given 
 alkali metal, increase with the atomic weights of the trivalent 
 metals ; and this increase becomes the more pronounced as we 
 pass from the caesium alums, through those of rubidium and 
 thallium, to the ammonium salts, successively. But the alums 
 of chromium, which stands between vanadium and iron in its 
 atomic weight, are much less soluble than those of either of 
 the latter metals. The sharp break in the curves, caused 
 by this behavior of the chromium salts, indicates that al- 
 though the method of plotting employed illustrates the gen- 
 eral gradation of the solubilities, the latter cannot truly be 
 regarded as a function of the atomic weights. The solubil- 
 ities, as determined, are as follows : 
 
 M m = Al. 
 
 v. 
 
 Cr. 
 
 Fe. 
 
 Cs 0.013 
 
 0.0204 
 
 0.0151 
 
 0.045 
 
 Rb 0.059 
 
 0.177 
 
 0.078 
 
 0.293 
 
 Tl 0.177 
 
 0.573 
 
 0.212 
 
 0.799 
 
 NH 4 0.387 
 
 1.210 
 
 0.407 
 
 1.659 
 
 Amer. Chem. Jour., xxvi, August, 1901. | See the preceding article. 
 
AND INORGANIC COMPOUNDS. 
 
 191 
 
 Cs M ra (S0 4 ), 
 
 - ^ ^^^^ 
 
 Atomic "Weights, 
 
 FIGURE I. 
 
 51 52 
 
 Fig. 1 reproduces these relations. For the sake of clear- 
 ness, the values are slightly increased along the a>axis. 
 
192 THE PERIODIC SYSTEM 
 
 The lines joining the vanadium and chromium alums of 
 each univalent metal should, in reality, be nearly perpendic- 
 ular to the z-axis. 
 
 A careful examination of this figure gives at the outset 
 one striking result, which, if it is also to be observed in 
 solubility charts of other series of homologous compounds, 
 may lead to a great advance in our knowledge of the rel- 
 ative influence of analogous elements upon their compounds. 
 
 The lines joining the solubility points of the successive univ- 
 alent metals with two given trivalent metals, have approximately 
 a common point of intersection. Thus, the line joining the sol- 
 ubility points of the caesium alums of vanadium and alumin- 
 ium, on prolongation, meets that of the thallium alums of 
 the same metals, in the point x= 16.25, y = 0.11 ; the unit on 
 the z-axis being the atomic unit, that on the ?/-axis one-hun- 
 dredth of a gram-molecule. The points of intersection of the 
 remaining corresponding lines vary only slightly from this, 
 and on either side of it. The intersection points of all the 
 aluminium-vanadium lines with one another follow : 
 
 Cs Rb. Rb TL Cs Tl. NH 4 Cs. NH 4 Rb. NH 4 Tl. 
 
 x = 18.1 16.30 16.25 16.00 15.70 15.10 
 y= 1.1 0.11 0.11 1.17 0.15 0.12 
 
 The following are the points of intersection of the vana- 
 dium-chromium lines: 
 
 Rb-Cs. Tl-Cs. Tl Rb. NH 4 Cs. NH 4 Rb. NH 4 -T1. 
 
 x = 52.99 52.6 52.65 52.6 52.6 52.53 
 y= 1.2 1.3 0.8 1.4 1.45 5.3 
 
 In both of these series the nearest approximation of the 
 points of intersection to a single point is seen to be in the 
 case of those lines which intersect at the greatest angle. 
 The most marked variations are that of the (AlCs VCs) and 
 (AlRb VRb) lines hi the first series, and that of the 
 (VTl-CrTl) and (VNH 4 -CrNH 4 ) lines in the second; that 
 is, where the lines in question are most nearly parallel, and 
 where, therefore, unavoidable error of experiment would pro- 
 
AND INORGANIC COMPOUNDS. 
 
 193 
 
 duce the greatest divergence. The various other lines, such as 
 those of (FeM'-CrM 1 ), (FeM-AlM 1 ), or (CrM'-AlM 1 ), 
 show a similar relation as regards their points of intersec- 
 tion, and it must be assumed, therefore, that the points rep- 
 resenting the solubilities stand in fixed mathematical relation to 
 one another. 
 
 This being so, the derivation of a general formula for the 
 calculation of the solubility of the sixteen alums in question 
 is a comparatively simple matter. The first step involved is 
 the empirical determination of the fixed points, of which the 
 observed points of intersection are approximations. Of 
 these, the three giving respectively the intersections of the 
 lines (VM'-AIM 1 ), (VM'-CrM 1 ) and (CrM'-FeM 1 ) are 
 required. 
 
 The points finally selected as most nearly satisfying all the 
 solubilities are as follows : 
 
 For (VM' - AIM') 
 
 Radii drawn from these points gave by intersection with 
 perpendiculars raised on the #-axis at points corresponding to 
 the atomic weights of the trivalent metals, the following solu- 
 bilities in gram-molecules. Under D is given the variation 
 from the observed solubilities. 
 
 
 M^Al 
 
 D. 
 
 V. 
 
 D. 
 
 Cr. 
 
 J>. 
 
 Fe. 
 
 D. 
 
 M'=Cs 
 Rb 
 Tl 
 
 NH 4 
 
 0.012 
 0.064 
 0.182 
 0.382 
 
 -0.001 
 +0.005 
 4-0.005 
 -0.006 
 
 0.0209 
 0.1878 
 0.5670 
 1.2090 
 
 +0.0005 
 +0.0108 
 -0.0060 
 -0.0010 
 
 0.0169 
 0.0734 
 0.2014 
 0.4182 
 
 +0.0018 
 -0.0046 
 -0.0106 
 +0.0112 
 
 0.0498 
 0.2776 
 0.7922 
 1.6643 
 
 +0.0048 
 -0.0169 
 -0.0068 
 +0.0016 
 
 Average enror, 
 
 +0.001 
 
 
 +0.001 
 
 
 -0.0005 
 
 
 -0.0043 
 
 13 
 
194 THE PERIODIC SYSTEM 
 
 As is seen, the agreement between the observed and calcu- 
 lated data is in the main very satisfactory. In more than half 
 the cases the difference does not exceed 0.005 gram-molecule. 
 The average molecular weight of an alum is about 300, and 
 the error represented by a difference of 0.001 gram-molecule 
 therefore means about 0.3 g. per liter. As the majority of 
 my determinations were made with quantities of solutions 
 containing about 3.0 g. of water, therefore, an error of this 
 magnitude would represent on the average less than a milli- 
 gram in weight ; or in the case of the ammonium alums, where 
 the dissolved salts were determined as sesquioxides, less than 
 0.5 milligram. A very slight variation in temperature, in the 
 case of the more soluble alums, introduces a still greater error. 
 At 30, one liter of water dissolves 0.467 gram-molecule 
 ammonium aluminium alum, and 0.495 gram-molecule ammo- 
 nium chromium alum. The increase in solubility for 5 is 
 therefore in the one case 0.080, in the other 0.088 gram-mole- 
 cule. Assuming that for so short an interval the solubility 
 is directly proportional to the temperature, a variation of 
 0.2 would cause an error of more than 0.003 gram-molecule. 
 The range of solubility of the alums is so great that errors of 
 such magnitude would have little influence upon the deter- 
 mination of the general solubility relations of the salts ; and 
 such relations were all that I had hoped to establish by the 
 work embodied in my last paper. If we take into account the 
 effect which the more or less extensive hydrolytic dissociation 
 of the sulphates of the trivalent metals would have upon the 
 solubility of the alums,* the variation between the solubilities, 
 calculated and observed, is in almost every case well within 
 the limit of permissible error. In the derivation of the con- 
 stants used in this paper, therefore, the calculated results will 
 be taken as correct. 
 
 The prolongation of the aluminium-vanadium lines and of 
 the vanadium-chromium lines to their respective points of 
 intersection yields a series of triangles which have a common 
 
 * This of course also applies to the formation of small quantities of basic 
 salts on evaporation of the solutions for analysis. 
 
AND INORGANIC COMPOUNDS. 
 
 195 
 
 base in the line connecting these points. A similar set of 
 triangles is given by the intersection of the vanadium-chro- 
 mium lines with those of chromium-iron. The resulting 
 figure, somewhat distorted in its proportions, is shown in 
 Fig. II. 
 
 The perpendiculars, AA1, BV, CCr, and DFe, and the 
 fixed points, P, Q, and S being given, the points of intersec- 
 tion of the radii from P, Q, and S with the perpendiculars B V, 
 CCr, and DFe, are necessarily determined by the position of 
 the points of intersection of the corresponding radii from P 
 with the perpendicular AA1. Now the points AI, B!, Ci, and 
 Dj represent the solubilities of the caesium alums of the suc- 
 cessive trivalent metals; A 2 , B 2 , C 2 , and D 2 , those of the 
 rubidium alums, etc. The line AiP makes a definite angle, 6, 
 
196 THE PERIODIC SYSTEM 
 
 with the a>axis, and the points B 1? Ci, DI, are accordingly 
 determined by the value of this angle. If we substitute rubid- 
 ium for caesium, 6 receives another value and B, C, and D 
 become B 2 , C 2 , D 2 . The effect of the substitution of any one 
 alkali metal for another in the alums of a trivalent metal is 
 therefore always measurable, directly or indirectly, by the 
 difference in the values of 6 peculiar to the alkali metals in 
 question. 
 
 The absolute values of the angle 6 are fixed by the relative 
 positions of the perpendiculars AA1 and B V upon the #-axis ; 
 but these do not affect its relative values; for if AA1, for 
 instance, be moved in either direction, the point P still re- 
 tains its y value, and therefore the relation 
 
 tan 0! A P A 2 A P 'A' 2 
 
 T- = -: 7 = -T 7-7- = const. 
 
 tan ApAi Ap>A \ 
 
 remains the same in all cases. is accordingly a variable 
 peculiar to the alkali metals in the compounds, and independ- 
 ent of the trivalent metals. The same is true of the variable 
 distance PA (/>), which depends directly upon the value of 
 0, in the sense, 
 
 p sin = AA P = const. 
 
 for each of the alkali metals, and without regard to the posi- 
 tion of the perpendiculars. 
 
 The effect of the trivalent metals upon the solubility is 
 also given by a variable which has a constant value for each. 
 As intercepts of two sides of a triangle by parallel lines, the 
 relations obtain : 
 
 . A 2 B 2 . A 3 B 8 
 " A 2 P A 8 P = 
 
 C 2 Q 
 
 "scT 
 
AND INORGANIC COMPOUNDS. 197 
 
 These constants, k, ki, and & 2 , are independent of the value 
 of the angle 0, and therefore of the univalent metals in the 
 alums. They are also independent of the positions of the 
 perpendiculars AA1, etc. If the relative positions of AA1 
 and BV be altered, for instance, until P becomes P' (Fig. Ill), 
 
 A 'B ' 
 
 the ratio between the new intercepts, A *, * , is still the same as 
 
 AT> AI r 
 
 l-Dl 7 
 
 or *. 
 We have, furthermore, the relation, 
 
 _ _ PA + AB _ P (k + 1) _ ^ . 
 
 A^ ~ A 2 A 3 PA 
 
 Similarly, for the vanadium and chromium alums, 
 dd C 2 C 8 dQ 1 
 
 BiB,, B 2 B 8 CiQ + kiCQ k t -f 1 
 D^ D 2 D 8 SO + 
 
 Now AiA 2 represents the difference in solubility of the 
 aluminium alums of rubidium and caesium ; CiC 2 , that of the 
 chromium alums, etc. If we call the difference in the solu- 
 bility of the alums of a given trivalent metal with two alkali 
 metals the " increment of solubility for the latter " (e. g., 
 Incr-Al^c,,), we arrive at the general law: 
 
 The ratio between the increments of solubility of the corre- 
 sponding alums of two trivalent metals for any two alkali 
 metals is constant. 
 
 The constant k indirectly represents, therefore, the effect 
 of the substitution of vanadium for aluminium in the alum 
 of a given univalent metal; A? lt that of chromium for vana- 
 dium, etc. And since the value of these constants is not 
 affected by the relative positions of the perpendiculars upon 
 the a>-axis, the atomic weights of the trivalent metals have 
 no determining influence upon them. The solubilities cannot, 
 therefore, be regarded as a function of these atomic weights. 
 
198 THE PERIODIC SYSTEM 
 
 The values of the constants k, etc., for the calculated solu- 
 bilities are as follows : 
 
 k = 2.2110 
 A* = 1.9608 
 k 2 = 3.0213 
 
 By substitution, we have, furthermore, for the ratio be- 
 tween the solubility increments of the alums of other pairs 
 of trivalent metals, 
 
 CiC 2 _ Incr.Cr m *_ m ' _ k + 1 _ 
 ~" Incr.Al m 2_ m ' k + 1 
 
 = 
 
 ^ 
 
 Incr.Al m 2_ m / k { + I 
 
 _ m / _ k 2 + 1 _ 
 
 ~ ~ 
 
 Owing to the large error introduced into the quotient of 
 the observed solubility increments by variations of a few 
 thousandths of a gram-molecule in the solubility determina- 
 tions, we can expect only an approximate agreement between 
 the above constants and their observed values. The latter, 
 which, together with the variations from the calculated values 
 (D), are given below, are therefore eminently satisfactory. 
 The agreement between them is so marked that there can be 
 no doubt as to the correctness of the law. 
 
 Incr.Al m2 _ m > Incr.O m2 _ m , 
 
 = 0.6611 
 
 . 
 
 Incr. V m 2_ m > Incr. V 
 
 m. m'. Observed. D. Observed. D. 
 
 NH 4 Cs 3.181 -0.030 0.3295 -0.0085 
 
 Tl Cs 3.378 +0.167 0.3561 +0.0184 
 
 Eb Cs 3.404 +0.190 0.4016 +0.0693 
 
 NH 4 Kb 3.149 -0.074 0.3185 -0.0192 
 
 Tl Eb 3.356 +0.145 0.3384 +0.0007 
 
 NH 4 Tl 3.033 0.178 0.3197 0.0316 
 
AND INORGANIC COMPOUNDS. 
 
 199 
 
 Incr.Fe 
 
 Incr.Cr m 2_ m ' 
 
 = 4.021 
 
 Incr.Or 
 
 Incr.Al 
 
 m 2 m' 
 
 m*. 
 
 m'. 
 
 Observed. 
 
 D. 
 
 NH 4 
 
 Cs 
 
 4.118 
 
 +0.097 
 
 Tl 
 
 Cs 
 
 3.829 
 
 -0.192 
 
 Kb 
 
 Cs 
 
 3.957 
 
 -0.064 
 
 NH 4 
 
 Kb 
 
 4.119 
 
 +0.098 
 
 Tl 
 
 Kb 
 
 3.770 
 
 -0.250 
 
 NH 4 
 
 Tl 
 
 4.410 
 
 +0.389 
 
 Observed. 
 
 1.021 
 1.200 
 1.152 
 1.003 
 1.127 
 0.927 
 
 = 1.084 
 
 D. 
 
 -0.063 
 +0.116 
 +0.068 
 -0.081 
 +0.043 
 -0.159 
 
 Incr.Fe m 2_ m ' 
 Incr.Al m 2_ m / 
 
 = 4.360 
 
 Incr.Fe 
 
 m 8 m' 
 
 m 2 . 
 
 m'. 
 
 Observed. 
 
 D. 
 
 NH 4 
 
 Cs 
 
 4.316 
 
 -0.044 
 
 Tl 
 
 Cs 
 
 4.537 
 
 +0.177 
 
 Kb 
 
 Cs 
 
 5.411 
 
 +1.051 
 
 NH 4 
 
 Kb 
 
 4.164 
 
 -0.196 
 
 Tl 
 
 Kb 
 
 4.200 
 
 -0.080 
 
 NH 4 
 
 Tl 
 
 4.095 
 
 -0.265 
 
 Incr.V m 2_ m / 
 
 Observed. 
 
 1.357 
 1.363 
 1.606 
 1.322 
 1.276 
 1.350 
 
 = 1.358 
 
 D. 
 
 -0.001 
 +0.005 
 +0.186 
 -0.031 
 -0.067 
 -0.008 
 
 The only marked deviations from the calculated constants 
 
 , ,, . 
 where both increments are 
 
 Incr.Cr Rb _ c _ 
 are in the case of 
 
 Tr 
 Incr.V Eb _c 8 
 
 so small that even the slightest error of experiment has a 
 great influence upon their ratio ; and in some of the ratios 
 obtained with rubidium ferric alum. The observed solubility 
 of this salt is probably somewhat high, owing to the readiness 
 with which it undergoes hydrolysis. The error thus intro- 
 duced, however, becomes very great only in the ratios 
 
 Incr.Fe Rb _ C8 
 
 In the ratios 
 
 Incr.M 
 
 its influence is 
 
 Incr.M Rb _ C8 
 hardly felt. 
 
 The remaining variables of the triangles PBQ and SCQ, 
 so far as their values are necessary for the derivation of a 
 general formula for the solubilities, are also to be expressed 
 in terms of 0, and &, k l9 etc. 
 
200 THE PERIODIC SYSTEM 
 
 If we call the angle of inclination of the base line PQ to 
 the 2>axis, a, then for the variable angle, BQP, or fa we 
 have the expression : 
 
 PB sin (Q - a) P (+l)sm(0-q) 
 
 Q * ~~ PQ - PB cos (0 a) ~~ PQ - P (k + 1) cos (0 - a) " 
 
 In this formula, a and PQ are constant. Substituting for 
 the latter and k their calculated values, we have, 
 
 _ p sin (0 q) __ ~ 
 Q *~ 11.37 -p cos (0-a) = 
 
 _ n BiP sin (0 - a) 
 
 Since BiQ = - -A- - *-> 
 
 Bin 9 
 
 the variable distance BQ becomes 
 
 p(A + 1) sin (0 - a) VTT^ 5 ,.. 
 -5- 
 
 As the intercepts BC = ki CQ and, therefore, 
 M = BC + ^, 
 
 KI 
 
 the relation which these intercepts bear to M are 
 
 BC = A^j- and CQ = r ^ T - 
 *i + 1 *! + 1 
 
 The base line SQ makes with the a>axis the constant 
 angle @ ; the variable angle BQS is therefore </> + ft a. 
 For the angle DSQ, or 2, we have, as in the case of the 
 angle <, 
 
 tanS- CQ sin (< + /? - a) 
 
 " SQ - CQ cos (<j> + ft - a) ' 
 
 The value of a is 1 5' 54", and that of A 26 33' 54". Ex- 
 
AND INORGANIC COMPOUNDS. 201 
 
 panding, and substituting for these and SQ their correspond- 
 ing values, and for CQ its equivalent, - = 
 
 KI -f- i. >s.youo 
 
 we have 
 
 Jf(# + 0.4763) 
 
 ~" 6.401 VI + IP M (1 - 0.4763 J?) * 
 
 The distance AA P = p sin 0. The solubility of the aluminium 
 alums is therefore expressed by the equation, 
 
 S^ = m + p sin (I) 
 
 m being the y-value of the point P. For the solubility of the 
 vanadium alums we have, 
 
 BB P = BB A + AA W (2) 
 
 and since BB A = k p sin 0, 
 
 by addition with (1), 
 
 S v = m + p sin + k p sin 0. 
 
 Passing in the same manner to the chromium alums, 
 CC P = BB P - BB C 
 
 = BB,-A1 sin (*-), 
 and, accordingly, 
 
 Sc, = m + p sin + k p sin , l 1 sin (<#> a). (3) 
 
 k\ ~r J- 
 
 The points D 1? D 2 e ^c., are expressed by 
 S Fe = m + DD P 
 
 = m + DD + CC P 
 
 in which DD C = CD sin (2 + /3). 
 
 The variable distance CD is given by the equation, 
 
 CQ sin (<ft + /? a) 
 
 sin 2 
 
202 THE PERIODIC SYSTEM 
 
 ft + /? a) 
 
 (&! + 1) sin S 
 Substituting this value for CD, we have 
 
 or 
 
 k M 
 S^ = m + p sin + k p sin ^ T sin (< a) + 
 
 1 T* 1 
 
 
 This also serves as a general formula for the solubility (S) 
 of any of the sixteen alums taken into consideration. For if 
 k 2 be made equal to zero, the last term falls away, leaving the 
 equation in the forms applying to the chromium alums. With 
 k z and k^ equal to zero, the solubilities of the vanadium alums 
 are obtained, and by eliminating the second term as well, 
 those of the aluminium salts. We thus have a general solu- 
 bility equation, all the terms of which can be referred to two 
 variables; the one of these, 0, applying to the one class of 
 elements in the compounds, the other, &, to the second class. 
 I believe that aside from the instance of addition properties, 
 which the solubilities certainly are not, this is the first case 
 in which a mathematical relation between the corresponding 
 properties of a class of compounds has been found capable of 
 expression. 
 
 Below I give the values of 6 and p for each of the alkali 
 metals, and the solubilities as calculated by the general 
 formula. With the latter, the column D contains the dif- 
 ference between the calculated and observed solubilities, and 
 D' the variation from the values serving as a basis for the 
 derivation of the constants. 
 
 9 p 
 
 Cs . . . . 2 6' 6" 10.945 
 
 Eb .... 27 11' 19" 12.254 
 
 Tl .... 57 56' 4 20.5406 
 
 73 45' 5" 38.943 
 
AND INORGANIC COMPOUNDS. 
 
 203 
 
 In the following the solubilities are expressed in hundredths 
 of a gram-molecule : 
 
 
 Al. 
 
 D. 
 
 D'. 
 
 J 
 II 
 
 M 
 
 D. 
 
 D'. 
 
 o- 
 II 
 
 J4 
 
 D. 
 
 D'. 
 
 w 
 
 * 
 
 Jf 
 
 D. 
 
 D'. 
 
 Cs 
 Rb 
 Tl 
 NH 4 
 
 1.192 
 6.38 
 18.21 
 38.18 
 
 -0.0108 
 +0.328 
 +0.510 
 +0.520 
 
 -0.008 
 -0.020 
 +0.010 
 -0.020 
 
 2.06 
 18.90 
 56.7 
 120.82 
 
 +0.02 
 +1.20 
 -0.60 
 -0.08 
 
 -0.03 
 +0.12 
 0.00 
 -0.07 
 
 1.671 
 7.34 
 20.13 
 41.82 
 
 +0.16 
 -0.36 
 -1.07 
 +1.12 
 
 -0.02 
 0.00 
 +0.01 
 0.00 
 
 4.82 
 27.76 
 79.69 
 165.26 
 
 +0.32 
 -1.63 
 -0.30 
 -0.64 
 
 -0.16 
 +0.06 
 +0.47 
 -1.17 
 
 In conclusion I may state that while the data now at hand 
 are too few for accurate analysis, such determinations as I 
 have made at temperatures other than 25 indicate that a 
 general solubility formula for all temperatures will not be 
 difficult of derivation. I have made determinations at 30 on 
 the caesium, rubidium, and ammonium alums of aluminium, 
 chromium, and iron ; and find that the law of the constant 
 ratio of the solubility increments holds good at that tempera- 
 ture. The value of the constants &, k^ and & 2 > however, vary 
 with the temperature. At 30 
 
 Incr.Cr m 2_ m ' 
 Incr.Al m 2_ m / 
 
 = 1.20, and 
 
 Incr.Fe m a- m 
 Incr.Cr m a_ m i 
 
 = 8.3 
 
 in approximate numbers. From the fact that this law obtains 
 at other temperatures, it necessarily follows that at these, as 
 at 25, the lines joining the solubility points have common 
 points of intersection. The same formula, therefore, with 
 other constants, obtains at different temperatures. The 
 mathematical relation between the solubilities at 25 cannot, 
 accordingly, be a matter of chance. 
 
 My thanks are due to Eugene Lamb Richards, Professor of 
 Mathematics in Yale University, for many helpful suggestions 
 embodied in this paper. 
 
PAPERS ON 
 DOUBLE HALOGEN SALTS 
 
ON SOME DOUBLE HALIDES OF SILVER AND 
 THE ALKALI METALS.* 
 
 BY H. L. WELLS AND H. L. WHEELER. 
 
 WITH THEIR CRYSTALLOGRAPHY. 
 
 BY S. L. PENEIELD. 
 
 DURING a systematic search for well-crystallized salts of the 
 type M'Hl.AgHl, f which we were anxious to obtain on ac- 
 count of their probable isomorphism with the alkaline tri- 
 halides, three well-defined compounds of another type, 
 2M / Hl.AgHl, were obtained. Our experience indicates that 
 these 2 : 1 salts are more easily prepared and crystallize better 
 than the 1 : 1 compounds. 
 
 The bodies to be described are 2CsCl.AgCl, 2RbI.AgI, and 
 2KI.AgI. Two of these are believed to be new salts; the 
 other, 2KI.AgI has been described by Boullay.J We have 
 not obtained a complete series of these compounds, for good 
 crystals could not be made of the other members, and, under 
 the circumstances, no products were analyzed except such as 
 could be measured. 
 
 The compounds are interesting from the fact that they do 
 not conform to Remsen's law concerning the composition of 
 double halides, for, contrary to this, they contain a number 
 of alkali-metal atoms which is greater than the number of 
 halogen atoms belonging to the silver. In his latest contribu- 
 tion to the subject, || Remsen states that the exceptions to his 
 law are "not more than three or four out of over four 
 
 * Amer. Jour. Sci., xliv, August, 1892. Amer. Chem. Jour., xi, 291. 
 t Ibid., Ill, xliii, 30 and 485. || Ibid., xiy, 87. 
 
 J Ann. Chim. Phys., II, xxiv, 377. 
 
208 SOME DOUBLE HALIDES OF SILVER 
 
 hundred." The work here described confirms the result of 
 Boullay, adds two more exceptions to the law, and points to 
 the existence of a greater number of compounds of the same 
 type. It may be mentioned that a considerable number of 
 other exceptions to this law have recently been established in 
 this laboratory and will soon be described. 
 
 Preparation and Properties. The salts are made by satu- 
 rating a very concentrated, hot solution of an alkaline halide 
 with the corresponding silver halide, filtering, cooling to 
 crystallization, and, if necessary, evaporating the mother-liquor 
 at ordinary temperatures. If the solutions are too dilute, in 
 some cases at least, the 1 : 1 salts are formed. The com- 
 pounds have little tendency to crystallize well, and many trials 
 are usually necessary in order to obtain satisfactory products. 
 The salts are all white. They are readily decomposed by 
 water. 
 
 Method of Analysis. The products analyzed were in the 
 form of crystals of such size that it was certain that they were 
 not mixed with other substances. In preparing them for 
 analysis the mother-liquor was removed rapidly and completely 
 by pressing them between smooth filter-papers, and great care 
 was taken to avoid any evaporation of the liquid which ad- 
 hered to them. The analyses were made by treating them 
 with a sufficient amount of water acidified with nitric acid 
 and weighing the silver halide thus separated. The filtrate 
 from this was used for determining the remaining halogen or 
 the alkali metal. 
 
 Tf nnn A Calculated for 
 
 2CsCl.AgCl. 
 
 Caesium 55.38 
 
 Silver 24.85 22.47 
 
 Chlorine 22.15 
 
 iPniin<i Calculated for 
 
 Found. 2RbI.AgI. 
 
 Bubidium . . . 25.05 25.91 
 
 Silver 17.32 16.36 
 
 Iodine .... 57.53 57.73 
 
 99.90 100.00 
 
AND THE ALKALI METALS. 
 
 209 
 
 Found. 
 
 Potassium 
 
 Silver 18.73 
 
 Iodine 
 
 Calculated for 
 2KI.AgI. 
 
 13.79 
 19.04 
 67.17 
 
 Crystallography. 
 
 The three salts are isomorphous and crystallize in the ortho- 
 rhombic system. The forms which were observed are : 
 
 a, 100, i-i 
 
 b, 010, i-i 
 
 m, 110, I 
 n, 120, i-2 
 
 d, 101, 1-t 
 x, 301, 3- 
 
 The axial ratios and some of the prominent angles are given 
 in the following tables, the fundamental measurements being 
 marked by an asterisk. The crystals did not yield very accu- 
 rate measurements. 
 
 U : ~b : c 
 
 0.244 
 0.236 
 0.234 
 
 d A d, 101 A foi 
 
 *28' 11' 
 *27' 12' 
 *27' 0' 
 
 2CsCl.AgCl. . . . 
 2EbI.AgI . . . . 
 2KI.AgI . . . . 
 
 m A m, 110 A 110 
 
 2CsCl.AgCl . . 88 18' 
 2EbI.AgI . . . *88 40' 
 2KI.AgI . . . *88 40' 
 
 0.971 : 1 
 0.977 : 1 
 0.977 : 1 
 
 n A n, 120 Al20 
 
 *54 29' 
 54 12' 
 54 12' 
 
 V 
 
 771 
 
 d 
 
 a 
 
 *~*^ 
 m 
 
 ^ 
 
 n 
 
 Fig. 1. 
 
 Fig. 2. 
 
 2CsCl.AgCl was made in minute prisms, less than a milli- 
 meter in diameter, having the habit shown in Fig. 1. The 
 measurements are only approximately correct. 
 
 14 
 
210 HALIDES OF SILVER AND THE ALKALI METALS. 
 
 Two crops of 2RbI. Agl were examined. One was like Fig. 1 
 in habit, the other in plates, Fig. 2. The crystals were nearly 
 10 mm. in length. On this salt a cleavage, parallel to a, was 
 observed; also, as small faces, the forms b and x, which are 
 not shown in the figures. In convergent polarized light an 
 obtuse bisectrix was seen, normal to a, the axial plane being 
 the brachy-pinacoid. 
 
 2KI.AgI was made in prismatic crystals, over 10 mm. in 
 length and having the habit and forms shown in Fig. 1. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1892. 
 
ON THE CAESIUM AND RUBIDIUM CHLOR- 
 AURATES AND BROMAURATES.* 
 
 BY H. L. WELLS AND H. L. WHEELER. 
 WITH THEIR CRYSTALLOGRAPHY. 
 BY S. L. PENFIELD. 
 
 A STUDY of the compounds to be described was undertaken 
 in the hope that some crystallographic analogy would exist be- 
 tween them and the alkaline pentahalides described in a pre- 
 vious article. f No such analogy has been found in spite of 
 the similarity of such formulae as CsCl.Cl 8 I and CsCl.Cl 3 Au, 
 but since some of these gold salts have never been described 
 and as they show some interesting relations among themselves, 
 our results are deemed worthy of publication. J 
 
 Th. Rosenbladt, in an article on the solubility of the chlor- 
 aurates, states that the caesium and rubidium salts lose their 
 water of crystallization almost completely when dried over sul- 
 phuric acid. He gives no statement of the amount of water, 
 but refers to his dissertation of 1872, which is inaccessible to 
 us. He mentions, however, that the crystals of both salts be- 
 long to the monoclinic system, so that it is probable that the 
 compounds he obtained were the ones that we have found to be 
 anhydrous. 
 
 The compounds that have been prepared are CsAuCl 4 , 
 2CsAuCl4.H 2 O, CsAuBr 4 , RbAuCL., and RbAuBr*. We 
 have attempted in each case to obtain bodies containing more 
 
 * Amer. Jour. Sci., xliv, August, 1892. 
 
 t Ibid., 42. 
 
 J The announcement by Professor Eemsen (Amer. Chem. Jour., xiv, 89), 
 that he and Mr. H. C. Jones proposed to examine the gold-rubidium halides, 
 was not made until after the work described in this article had been 
 completed. 
 
 Berichte, xix, 2535. 
 
212 ON THE CESIUM AND RUBIDIUM 
 
 caesium and rubidium, but no evidence of their existence has 
 been found. 
 
 An investigation of the corresponding iodine compounds 
 was also undertaken, but, on account of the instability of 
 auric iodide, we did not obtain any pure or well-crystallized 
 products. 
 
 Preparation. The salts are so insoluble that they form 
 precipitates when moderately concentrated solutions of the 
 component salts are mixed, and the products are readily re- 
 crystallized from water or from the mother-liquors. It is 
 usually immaterial whether the solutions are neutral or acid 
 or whether the gold or alkaline halide is in excess, but the salt 
 2CsAuCl 4 .H 2 O requires special conditions for its preparation, 
 for it is apparently formed only when a large excess of gold 
 chloride is present and when the solution does not contain 
 much free acid. We have used four atoms of gold to one of 
 csesium in making this salt, but it usually requires repeated tri- 
 als under these conditions before it is obtained free from the 
 anhydrous compound. The two salts are however so distinct 
 in form that there is no difficulty in distinguishing them. 
 
 Properties. The color of CsAuCl 4 and of 2CsAuCl 4 .H 2 O 
 is golden-yellow ; RbAuCl 4 is yellowish-red ; the two bromides 
 are black, but give a dark red powder. 
 
 All the salts are sparingly soluble in water, especially when 
 cold, and the csesium compounds are less soluble than the ru- 
 bidium. All of them are only slightly soluble in alcohol and 
 insoluble in ether. 
 
 Methods of Analysis. The crystals were prepared for anal- 
 ysis by quickly pressing them between smooth filter-papers 
 and finally allowing them to become air-dry. The hydrous 
 caesium chloraurate, however, loses its water and becomes 
 opaque on exposure. It was therefore dried as rapidly and 
 thoroughly as possible on paper, and was put into a weighing- 
 tube as soon as some of the fragments began to lose their 
 transparency. 
 
 Gold was determined by precipitation with ammonium oxa- 
 late or with sulphurous acid. The filtrate from the metallic 
 
CHLORAURATES AND BROMAURATES. 
 
 213 
 
 gold was used either to determine the alkali metal as normal 
 sulphate or the halogen by the usual gravimetric method. 
 Water was determined by the method used in the combustion 
 of organic compounds, the halogens being held back by a mix- 
 ture of lead chromate and lead oxide. The absence of water 
 in the anhydrous compounds was established by the use of the 
 same process. 
 
 Caesium 
 Gold . 
 Chlorine 
 
 Caesium 
 Gold . 
 Chlorine 
 Water 
 
 Caesium 
 
 Gold 
 
 Bromine 
 
 Rubidium 
 Gold . . 
 Chlorine 
 
 Rubidium 
 Gold . . 
 Bromine 
 
 Pound. 
 
 27.23 
 40.23 
 29.07 
 2.32 
 98.86 
 
 2.37< 
 
 2.20* 
 
 Found. 
 
 20.73 . . . 
 30.32 30.26 
 49.31 
 
 100.36 
 
 FouncL 
 
 45.53 
 32.98 
 
 Found. 
 
 32.54 
 
 Calculated for 
 CsAuCV 
 
 28.16 
 41.77 
 30.06 
 
 Calculated for 
 2CsAuCl 4 .H 2 O. 
 
 27.63 
 
 40.99 
 
 29.50 
 
 1.87 
 
 Calculated for 
 CsAuBr*. 
 
 20.45 
 30.34 
 49.21 
 
 Calculated for 
 RbAuCl*. 
 
 20.14 
 46.46 
 33.40 
 
 Calculated for 
 RbAuBr 4 . 
 
 14.18 
 32.73 
 53.08 
 
 * From a separate product. 
 
214 
 
 ON THE CAESIUM AND RUBIDIUM 
 
 Crystallography. 
 
 The crystallization of CsAuCU, CsAuBr 4 , RbAuCl 4 , and 
 RbAuBr 4 is monoclinic. The four salts form an isomorphous 
 group and are identical in crystalline habit. The forms which 
 have been observed on them are : 
 
 c, 001, 
 m, 110, 1 
 
 2. 
 
 d, 021, 2-i 
 
 e, 201, 2-1 
 
 /K /r\ 
 
 771 
 
 m 
 
 The crystals are prismatic and are usually terminated by e, 
 Fig. 1. When other faces are present, they are always small, as 
 represented in Fig. 2. The pyramid p, which is not shown in 
 the figure, frequently occurs as a small face, replacing the 
 edge between d and e. Among the crystals of CsAuBr 4 several 
 twins were observed, having p, 111 as the twinning plane, Fig. 
 3, while Fig. 4 represents a crystal of RbAuBr 4 twinned about 
 e, 201. The letters belonging to the parts in twin position are 
 underlined. Both kinds of twins are abnormally developed, as 
 represented in the figures. In all four compounds the cleav- 
 age is perfect, parallel to the base. 
 
 The rubidium salts, being the most soluble, form readily in 
 large crystals, several centimeters in length. The chloride, 
 especially, yielded magnificent crystals, which frequently were 
 only limited in length by the size of the vessel and volume of 
 the solution containing them. The caesium salts are less solu- 
 
CHLORAURATES AND BROMAURATES. 
 
 215 
 
 ble and were made in small prisms, seldom over 5 mm. in length. 
 The crystals were frequently hollow or cavernous at the ex- 
 tremities; this was especially true of the two bromides. The 
 faces, for the most part, gave excellent reflections of the signal 
 on the goniometer. 
 
 The axial ratios are as follows: 
 
 CsAuCl 4 
 
 I : c = 1.1255 : 1 : 0.7228 
 P = 71 36' 
 
 BbAuCU 
 
 I : e = 1.1954 : 1 : 0.7385 
 B = 75 32' 
 
 In the following tables the angles which were chosen as 
 fundamental are marked by an asterisk: 
 
 CsAuBr 4 
 
 a : I : c = 1.1359 : 1 : 0.7411 
 P = 70 24J' 
 
 RbAuBr 4 
 
 a:5:c: = 1.1951: 1:0.7256 
 
 = 76 
 
 CsAuBr 4 
 
 110 
 
 Measured. 
 
 m A m, A 1TO = *93 46' 
 m A c, 110 A 001 = 77 36' 
 m A d, 110 A 021 = 44 6' 
 d A p, 021 A Til = ... 
 d A e, 021 A 201 = *75 17' 
 m A e, TTO A 201 = *60 36' 
 c A e, 001 A 201 = 64 20' 
 
 Calculated. 
 
 77 32' 
 44 7' 
 
 Measured. 
 
 *93 53' 
 
 Calculated. 
 
 *76 46' 
 
 ... 
 
 43 23' 
 
 43 20' 
 
 32 23' 
 
 75 31' 
 
 32 40J' 
 75 59' 
 
 *60 41' 
 
 . 
 
 64 18' 
 
 m A m, Ke-entrant angle of twin, 
 
 BbAuCl* 
 
 27 58' 27 58' 
 
 RbAuBr 4 
 
 110 
 
 Measured. 
 
 m A m 9 JLJLU A 1TO = *98 21' 
 m A c, 110 A 001 = *80 36' 
 m A d 9 110 A 021 = ... 
 d* p 9 021 A Til = ... 
 d A e, 021 A 201 = ... 
 m A e, HO A 201 = *62 12' 
 c A c, 001 A 201 = 60 4' 
 d A d, 021 A 021 = 110 20' 
 
 Calculated. 
 
 59 59' 
 110 4i 
 
 Measured. 
 
 Calculated. 
 
 *98 40' 
 
 ... 
 
 *81 30' 
 
 . . . 
 
 44 57' 
 
 45 12J' 
 
 31 26' 
 
 31 35' 
 
 72 28' 
 
 72 26' 
 
 62 9' 
 
 62 21*' 
 
 *109 26' 
 
 
 55 42' 
 
 55 17' 
 
 m A m, Re-entrant angle of twin, 
 
 In their axial ratios the two caesium salts are very similar, 
 as are also the two rubidium salts, while the rubidium com- 
 
216 
 
 ON THE CESIUM AND RUBIDIUM 
 
 pounds differ considerably from those of caesium, especially in 
 the relation of d to the other axes and in the angles /3. It is 
 therefore evident that the replacement of one metal by another 
 in these salts has a considerable influence upon their form, 
 whereas, as we have shown, such a replacement in the caesium 
 and rubidium trihalides has little or no effect. There seems 
 to be no regularity in the influence of the replacement of 
 chlorine by bromine in these gold salts, for in the caesium 
 compounds the chloride has a slightly shorter axis 6 and a 
 greater angle fi than the bromide, while in the rubidium salts 
 exactly the reverse is true in both cases. This unexpected 
 relation between the chlorides and bromides has been con- 
 firmed by repeating the measurements, especially of the angle 
 m A c, using both crystal and cleavage faces. It is certain that 
 this angle is about a degree greater with the chloride than with 
 the bromide in the caesium salts, while in the rubidium com- 
 pounds it is about a degree less. 
 
 The crystallization of 2CsAuGl 4 .H 2 O is ortho- 
 rhombic. This salt was repeatedly made, but 
 only one crop of crystals was obtained which 
 was suitable for measurement. These were thin 
 plates, having the habit shown in Fig. 5. They 
 were not over 5 mm. in length and were only 
 a fraction of a millimeter thick. On removal 
 from the mother-liquor or from a moist atmos- 
 phere, the transparent plates rapidly became 
 opaque and the faces lost their lustre, so that only approximate 
 measurements could be obtained. 
 The forms which were observed are : 
 
 a, 100, i+ 
 
 b, 010, Vt 
 
 110, 1 
 120, i-2 
 
 d, 101, 1-t 
 
 The axial ratio is as follows : 
 
 :5:c = 0.625:l:0.24 
 
CHLORAURATES AND BROMAURATES. 217 
 
 The following measurements were made : 
 
 a A m, 100 A 110 = about 32 a A ft, 100 A 010 = about 90 
 a A n, 100 A 120 = about 51 d A d, 101 A T01 = about 42 
 
 Under the polarizing microscope the crystals show parallel 
 extinction and, in convergent light, an acute bisectrix normal 
 to #, 100. The plane of the optical axes is the base. The 
 divergence of the axes is large, the hyperbolae opening out 
 beyond the field of the microscope. The axes of elasticity 
 are: 
 
 d t, I a, c = fc. 
 
 The double refraction is therefore positive. 
 
 The change which the crystals undergo when exposed to 
 dry air is a molecular rearrangement, accompanied by loss of 
 water and probably a change to the anhydrous salt which was 
 described above. This rearrangement is a beautiful sight 
 when studied with the microscope in polarized light. The 
 change commences a few minutes after the crystals are re- 
 moved from the mother-liquor, and in less than ten minutes 
 has usually advanced to such an extent that the crystals are 
 no longer transparent. The crystals at first show a uniform 
 action on polarized light ; then from different parts of the 
 surface the rearrangement, which is marked by aggregate 
 pokrization, commences. It advances, shooting out in va- 
 rious directions in a manner resembling the growth of ammo- 
 nium chloride crystals under the microscope, until the whole 
 field is covered and light is finally no longer transmitted. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 April, 1892. 
 
ON THE CAESIUM-MERCURIC HALIDES.* 
 
 BY H. L. WELLS. 
 
 IT is to be expected that more complete series of double 
 halides can be made with caesium than with the other alkali 
 metals, because it is the extreme member of the potassium 
 group and the most electro-positive element known, and be- 
 cause caesium double salts in general are less soluble than the 
 corresponding compounds of the other alkali metals. A thor- 
 ough study of these compounds seems desirable, since very 
 little work has been done in this direction, and therefore the 
 present investigation of the caesium-mercuric chlorides, bro- 
 mides, and iodides has been undertaken. 
 
 The following is a complete list of the previously described 
 mercuric double halides containing the alkali metals and am- 
 monium, as far as I have been able to find them : 
 
 Na 2 HgCl 4 NH 4 HgCl 8 RbHg 2 Cl 5 
 
 Kb 2 HgCl, EbHgCl 8 KHg 2 Cl 6 .2H 2 
 
 Cs 2 HgCl 4 KHgBr 8 
 
 (NH 4 ) 2 HgBr 4 NH 4 HgCl 8 .pI 2 
 
 K 2 HgBr 4 KHg01 8 .H 2 (NH 4 ) 2 Hg 8 Cl 8 .4H 2 
 
 Na 2 HgI 4 KHgBr 8 .H 2 
 
 K 2 HgI 4 NaHgCl 8 .HH 2 
 
 (NH 4 ) 2 HgCl 4 .H 2 NH 4 HgI 3 .l^H 2 (NH 4 ) 2 Hg 9 Cl 20 
 
 K 2 HgCl 4 .H 2 KHgI 8 .liH 2 
 
 Kb 2 HgCl 4 .2H 2 
 
 (NH 4 ) 2 HgI 4 .3H 2 
 
 The greater number of these arrange themselves into two 
 types with varying water of crystallization or with none. 
 There are two compounds of a third type, while the two re- 
 
 * Amer. Jour. Sci., xliv, September, 1892. 
 
THE CESIUM-MERCURIC HALIDES. 219 
 
 maining, more complicated salts, stand alone. The last two 
 were described by Holmes.* 
 
 An effort has been made to make the examination of the 
 cesium-mercuric salts very complete, but it is not safe to say 
 that every possible compound has been prepared, for negative 
 results are uncertain. It will not be necessary to describe the 
 unsuccessful experiments where mixtures or uncertain prod- 
 ucts were obtained. It is sufficient to say that other double 
 halides were repeatedly looked for in every direction, and 
 every indication of a new salt was followed up until a homo- 
 geneous product was obtained and analyzed. 
 
 The following table gives a list of the salts that are to be 
 described. One of them, Cs 2 HgCl 4 , has already been prepared 
 by Godeffroy.f 
 
 i. n. ra. 
 
 Cs 3 HgCl 5 Cs 2 HgCl 4 CsHgClst 
 Cs 3 HgBr 5 Cs 2 HgBr 4 CsHgBr 8 | 
 Cs 3 HgI 5 Cs 2 HgI 4 CsHgI 3 
 Cs 3 HgCl 3 Br 2 Cs 2 HgCl 2 Br 2 CsHgClBr,* 
 Cs 8 HgBr 3 I a Cs 2 HgBr 2 I 2 1: CsHgBrl, 
 Cs 2 HgCl 2 I 2 
 
 IV. V. VL 
 
 CsHg 2 Cl 6 CsHg 8 Cl u 
 
 CsHg 2 Br fi .... 
 
 Cs 2 Hg 3 I 8 CsHg 2 I 5 
 
 CsHg 2 ClBr 4 CsHg 6 ClBr 10 
 
 These salts confirm the composition of all the previously 
 known alkaline-mercuric halides, as given in the preceding 
 table, except the single compound (NH 4 ) 2 Hg 9 Cl 2 o. It is ex- 
 tremely probable, however, that the correct formula for this is 
 NH 4 Hg 6 Cln, for Holmes obtained results slightly lower than 
 his theory in his ammonium determinations, and it would be 
 scarcely possible to distinguish between the two formulas by 
 analysis, as will be seen from the following numbers : 
 
 * Chem. News, v, 351. t Berichte, viii, 9. 
 
 J These compounds are dimorphous. 
 
220 THE CAESIUM-MERCURIC HALIDES. 
 
 Calculated for Calculated for ru, 
 
 (N^) a Hg 9 Cl M . NH 4 Hg 6 C IV Differences. 
 
 Mercury ..... 70.70 71.00 0.30 
 
 Ammonium .... 1.41 1.27 0.14 
 
 The differences between the amounts of mercury and caesium 
 for the corresponding formulas are 0.80 and 0.85, so that it 
 is evident that the csesium compound furnishes a far better 
 means of determining the composition of the salts. 
 
 The first type, Cs 8 HgHl 6 , is a new one. These compounds 
 are interesting as exceptions to Remsen's law concerning the 
 composition of double halides.* 
 
 The salt Cs 2 Hg 3 I 8 , although standing alone among the cse- 
 sium compounds, is a very well characterized body, and the 
 compound (NH 4 ) 2 Hg 8 Cl8.4H 2 O, made by Holmes, belongs to 
 the same type. 
 
 The results of the work on the caesium-mercuric salts fulfil 
 the expectations concerning the value of csesium as a means of 
 studying alkaline double halides, for all the previously discov- 
 ered types have been made with this metal, and one besides 
 that had never been discovered. 
 
 Preparation. 
 
 The compounds were made by dissolving mercuric halides 
 in hot solutions of csesium halides and cooling, or in some 
 cases evaporating at ordinary temperatures, to crystallization. 
 The relative amounts of the two halides and the dilution both 
 have an important influence in determining the salt produced. 
 In most cases dilution with water is equivalent to the addition 
 of mercury, while concentration produces the same effect as 
 the addition of a csesium halide. It has been noticed, where 
 more than one salt is deposited from a solution by cooling, 
 that the salts with more mercuric halides are formed first. 
 This shows that cooling a solution may be equivalent to the 
 addition of csesium. 
 
 There are only a few of the salts that can be recrystallized 
 unchanged from water, most of them requiring the presence of 
 
 * Amer. Chem. Jour., xi, 296 ; xiv, 86. 
 
THE CAESIUM-MERCURIC HALIDES. 221 
 
 an excess of caesium halide, or in two or three cases mercuric 
 halide, for their formation. Crystallization from water can 
 therefore often be used for preparing one salt from another. 
 
 All the compounds were made with solutions of the normal 
 salts without the use of acids. Some of them have been made 
 with alcoholic solutions, but this solvent has not been found 
 to possess any advantages except for preparing CsHg 2 I 6 . 
 
 Analytical Methods. 
 
 The salts were always carefully examined to be sure that 
 they were not mixtures. Many mixed crops of crystals were 
 obtained, but I am confident that the products analyzed were 
 pure. The crystals for analysis were always quickly and thor- 
 oughly freed from the mother-liquor by pressing repeatedly 
 between smooth filter-papers, and at the same time they were 
 crushed to remove included liquid. During this drying pro- 
 cess the substances were exposed to the air as little as possible 
 to avoid any evaporation of the adhering liquid before its 
 removal. After the products had been dried as thoroughly as 
 possible in this way, they were usually exposed to the air for 
 an hour or two to remove the last traces of moisture, but this 
 was not done in a few cases where I wished to be certain that 
 no easily lost water of crystallization was present. 
 
 Portions of about one gram of substance were usually taken 
 for analysis. In no case was the analysis hampered from kck 
 of material. The chlorides and bromides were readily dis- 
 solved in water, but it was necessary in analyzing the iodine 
 compounds to dissolve them in water containing alcohol. 
 Mercury was invariably determined as sulphide, the precipitate 
 being collected, dried at 100, and weighed on an asbestos filter 
 in a Gooch crucible. Caesium was usually determined in the 
 filtrate from the mercuric sulphide and was always weighed as 
 sulphate. In this operation the excess of sulphuric acid was 
 removed by ignition in a current of air containing ammonia, 
 as suggested by Kriiss for potassium sulphate. In some cases 
 where caesium alone was to be determined, the substance was 
 weighed out directly into a platinum crucible, sulphuric acid 
 
222 THE CAESIUM-MERCURIC HALIDES. 
 
 was added, the excess of this and the mercury were removed 
 by evaporation and heating, and normal caesium sulphate was 
 weighed. The halogens were invariably determined in sep- 
 arate portions and were weighed as silver salts. In the cases 
 where two were present, they were determined by heating the 
 mixed silver halides to constant weight in chlorine. 
 
 The Double Chlorides. 
 
 These are all white in color and are permanent when ex- 
 posed to the air. On recrystallizing from water all of them 
 finally yield CsHgCl 8 . 
 
 Cs&Hg Cls is made by dissolving a comparatively small quan- 
 tity of mercuric chloride in a nearly saturated csesium chloride 
 solution. It is deposited on cooling, but the best crystals are 
 obtained by spontaneous evaporation. If too much of the 
 mercuric compound is added or if too much water is present, 
 other double salts or mixed products will be obtained. On the 
 other hand, if too little mercuric chloride is present, csesium 
 chloride crystallizes out. The limits of the conditions under 
 which it is formed are narrow, but by repeated trials, with 
 slight variations suggested by previous results, a pure product 
 is readily obtained. It forms slender, radiating prisms which 
 are easily distinguished from the compounds with which it is 
 liable to be mixed. 
 
 The following analysis was made of a sample which was 
 rapidly dried on paper, but not air-dried. The small amount 
 of water found was probably simply moisture. It was deter- 
 mined by direct weighing in a calcium-chloride tube. 
 
 ,, , Calculated for 
 
 Cs 3 HgCl 6 . 
 
 Caesium . . . 51.15 51.38 
 
 Mercury . . . 24.84 25.76 
 
 Chlorine . . . 21.79 22.86 
 
 Water . . . 1.69 0.00 
 
 99.47 100.00 
 
 CsJIgCli is produced, by cooling a hot solution, when a 
 little more mercuric chloride or water is used than in the case 
 
THE CESIUM-MERCURIC HALIDES. 223 
 
 of the last salt. The conditions for its formation are narrow. 
 It forms large but usually very thin plates, which are readily 
 distinguished from the other double chlorides. A sample was 
 dried on paper for analysis. 
 
 Osium . . . 44.06 43.75 
 
 Mercury ...... 32.90 
 
 Chlorine . . . 22.87 23.35 
 
 Water .... 0.52 0.00 
 
 100.00 
 
 is dimorphous, forming, according to circumstan- 
 ces, cubic or orthorhombic crystals. The cubic form is pro- 
 duced under widely varying conditions by cooling dilute 
 aqueous solutions, when caesium chloride is considerably in 
 excess. The orthorhombic form is deposited when caesium 
 chloride is not in great excess, and by one or more recrystal- 
 lizations from water of all the double chlorides. This form 
 can be recrystallized from water indefinitely. 
 
 The compound is practically insoluble in absolute alcohol, 
 but it dissolves in alcohol diluted with about one-third of its 
 volume of water, and it is remarkable that the cubic form is 
 deposited from such a solution on cooling. 
 
 The cubes often form peculiar aggregates, apparently of a 
 pyramidal shape. The orthorhombic crystals are very brilliant 
 and highly modified, usually forming groups of spear-shaped 
 individuals joined end to end. 
 
 Three samples were analyzed: A, cubes simply dried on 
 paper; B cubes from alcohol; C, orthorhombic crystals, air- 
 dry. 
 
 Calculated for 
 T CsHgCl,. 
 
 Caesium . . . 30.29 30.26 29.92 30.26 
 
 Mercury . . . 44.80 . . . 45.63 45.51 
 
 Chlorine . . . 23.40 . . . 24.03 24.23 
 
 Water . . . 1.42 . . . J _ LI _ 0.00 
 
 99.91 99.58 10OOO 
 
224 THE CESIUM-MERCURIC HALIDES. 
 
 Since the orthorhombic form of this compound is not de- 
 composed by water, its solubility could be determined. This 
 was done by analyzing the mother-liquor from a third recrys- 
 tallization at about 17. Of this solution, 100 parts contained 
 0.4255 parts of caesium, corresponding to 1.406 parts of 
 CsHgCls. 
 
 CsHgtCh was made by dissolving 24 g. of CsHgClg and 16 g. 
 of HgCl 2 (a little more than one molecule of the latter) in 
 about 150 c. c. of hot water and cooling. A large crop of 
 needles was obtained which were undoubtedly homogeneous. 
 
 Analysis gave *%$&?* 
 
 Cesium . . . 18.13 18.72 
 
 Mercury . . . 56.32 56.30 
 
 Chlorine . . . 24.68 24.98 
 
 99.13 100.00 
 
 The salt is not very readily decomposed by water, but by 
 repeated recrystallization the orthorhombic form of CsHgCl 3 
 is obtained. 
 
 CsHg&Cln was prepared by making a nearly saturated solu- 
 tion of 12.5 g. of HgCsCla and 38.5 g. of HgCl 2 (about one 
 molecule of CsCl to six of HgCl 2 ) in boiling water and cool- 
 ing. The compound was obtained in prisms, so well formed 
 that there was no doubt about their homogeneity. Two crops 
 were analyzed. 
 
 ,, , Calculated for 
 
 CsHg 6 Cl u . 
 
 Caesium . . 8.68 8.51 8.73 
 
 Mercury . . 65.59 . . . 65.64 
 
 Chlorine . . 24.97 . . . 25.63 
 
 99.24 100.00 
 
 A single recrystallization of this salt from water gave a 
 mixed crop of crystals, and this, on repeating the operation, 
 gave CsHg 2 Cl 6 , still containing a little of the original com- 
 pound. This last crop was analyzed. 
 
 Found. 
 
 Calculated for 
 CsHgjCl 5 . 
 
 Caesium 15.57 18.72 
 
THE CESIUM-MERCURIC HALIDES. 225 
 
 The Double Bromides. 
 
 All of these salts are white, or nearly so, except CsHgBr 8 , 
 which has a lemon-yellow color. This color is remarkable, 
 since CsBr and HgBr 2 are both pure white. 
 
 All of the double bromides yield CsHg 2 Br 5 on recrystallizing 
 them one or more times from water. It is to be noticed that 
 this salt belongs to a different type from the double chloride 
 which is stable with water, but if alcohol is used for recrys- 
 tallizing this bromide, the salt corresponding to the chloride 
 just mentioned is deposited. 
 
 CszHgBr^ The preparation of this salt is exactly analogous 
 to that of the corresponding chloride, and it has the same 
 appearance. 
 
 Found Calculated for 
 
 Cs s HgBr B . 
 
 Caesium . . . 39.83 39.94 
 
 Mercury . . . 19.50 20.02 
 
 Bromine . . . 39.60 40.04 
 
 98.93 100.00 
 
 CszHgBr^ is prepared similarly to the chloride, but the 
 limits of the conditions under which it is formed are much 
 wider. Like the chloride it usually forms very thin plates, 
 but they can sometimes be produced of sufficient thickness for 
 measurement. Three separate crops, made under considerably 
 different conditions, were analyzed. 
 
 Caesium 
 
 Mercury 
 
 Bromine 
 
 
 Found. 
 
 
 Calculated for 
 Cs 2 HgBr 4 . 
 
 33.84 
 
 33.84 
 
 34.43 
 
 33.69 
 
 25.68 
 
 25.11 
 
 25.45 
 
 25.45 
 
 40.48 
 
 40.40 
 
 40.52 
 
 40.71 
 
 100.00 
 
 99.94 
 
 99.66 
 
 100.00 
 
 . This compound is dimorphous, but while one 
 form is cubic, like one of the chlorides, the other is mono- 
 clinic and has no apparent relation to the orthorhombic 
 chloride. Just as in the case of the chlorides, the cubic form 
 is produced when an excess of the caesium halide is present, 
 while the second form is deposited when this excess is not as 
 
 15 
 
226 THE CAESIUM-MERCURIC HALIDES. 
 
 great. Unlike the corresponding chloride, the second form of 
 the bromide is decomposed by reciystallization from water, the 
 salt CsHg 2 Br s being formed, but, as will be noticed beyond, 
 the opposite transformation can be produced by recrystallizing 
 the last-mentioned salt from alcohol. The limits of formation 
 of the cubic salt are wide, but it is difficult to produce the 
 other form in a pure state, and it is possible that the mono- 
 clinic crystals analyzed were mixed with a small quantity of 
 the cubes. 
 
 Found. Calculated for 
 
 Cubic^ Monoclink CsHgBr 9 . 
 
 Cesium . . . 23.18 22.89 23.21 
 
 Mercury . . . 34.95 35.54 34.90 
 
 Bromine . . . 41.70 41.63 41.89 
 
 99.83 100.06 100.00 
 
 CsHg^Br^ The recrystallization of any of the other 
 double bromides from water produces this salt, and it can be 
 recrystallized indefinitely without decomposition. It forms 
 very small, thin plates which have a very faint tinge of yellow. 
 By spontaneous evaporation of a mother-liquor from a recrys- 
 tallization of this salt somewhat larger crystals were formed. 
 Three separate crops were analyzed. 
 
 Found. Calculated for 
 
 , * s CsHg 2 Br 6 . 
 
 jium . . . 14.60 14.69 13.24 14.26 
 
 Mercury . . . 42.71 ...... 42.87 
 
 Bromine . . . 42.55 ...... 42.87 
 
 99.86 100.00 
 
 The mother-liquor from a third recrystallization from water 
 at about 16 was found to contain 0.1151 per cent of caesium, 
 corresponding to solubility of 0.807 parts of CsHg 2 Br 5 in 100 
 parts of the solution. The salt dissolves rather sparingly in 
 hot, strong alcohol and, on cooling this solution, the compound 
 CsHgBr 8 separates out. 
 
 Caesium . . . 22.68 23.21 
 
THE CAESIUM-MERCURIC HALIDES. 227 
 
 The crystals thus obtained were not large enough to measure, 
 but it was probable, from microscopic examination, that they 
 were the monoclinic form of this compound. This is interest- 
 ing from the fact that it is the cubic form of CsHgClj which 
 crystallizes from alcoholic solutions. 
 
 No satisfactory crops of crystals were obtained from solu- 
 tions made with CsHg 2 Cl 6 and HgBr 2 , together. 
 
 The Double Iodides. 
 
 These salts are all yellow, CsHg 2 I 2 and Cs 2 Hg 8 I 8 having a 
 color nearly like that of normal potassium chromate, while the 
 others become paler as the csesium chloride increases. All of 
 them are decomposed by water, forming compounds containing 
 more mercuric iodide than the original salt, or, at last, mercuric 
 iodide itself. It is therefore possible to take any one of these 
 double salts, and, by recrystallizing from water and evaporating 
 the resulting solutions, to prepare the complete series of five 
 double iodides, as well as the component simple iodides, with- 
 out the use of any new material. It is noticeable that the 
 iodides differ from the chlorides and bromides in not including 
 a salt that can be recrystallized continually from water. This 
 peculiarity is doubtless due to the comparative insolubility of 
 mercuric iodide. In most cases the analyses of the salts con- 
 taining iodine show an excess of mercury and a deficiency of 
 the halogen (or halogens). It is not known whether this was 
 due to some impurity in the salts or to analytical errors. It is 
 not considered probable that inaccuracies in the analyses could 
 have caused so much variation from theory, for the methods 
 used were the same as for the chlorides and bromides, except 
 that alcohol was used as a solvent, and, while halogens and 
 mercury were always determined in separate portions, the 
 summations of the analyses were usually satisfactory. 
 
 C8 s HgI 5 . This salt, like the corresponding chloride and 
 bromide, requires for its preparation a very concentrated solu- 
 tion of the csesium halide containing a relatively small amount 
 of the mercuric compound. It crystallizes well and may be 
 
228 THE CAESIUM-MERCURIC HALIDES. 
 
 obtained either by cooling or spontaneous evaporation. The 
 crystals form peculiar, steep pyramids. 
 
 Vn * Calculated for 
 
 Found ' 
 
 Caesium . . . 33.02 32.33 
 
 Mercury . . . 16.33 16.21 
 
 Iodine .... 50.42 51.46 
 
 99.77 100.00 
 
 Its specific gravity, taken in benzol, was found to be 4.605. 
 
 When this salt is dissolved in a small quantity of hot water, 
 the compound Cs 2 HgI 4 crystallizes out on cooling, but with a 
 larger quantity of water everything remains in solution. 
 
 Cs z HgIi. This salt is produced under wide limits of condi- 
 tions by cooling solutions of the component salts when caesium 
 iodide is in excess. The monoclinic crystals vary in habit, 
 forming long prisms, nearly square plates or intermediate 
 forms. They are often obtained of very large size, sometimes 
 extending completely across the bottom of the vessel contain- 
 ing the solution and turning upward at the ends besides. 
 
 Caesium 
 Mercury 
 Iodine . 
 
 Two determinations of the specific gravity, taken in benzol, 
 gave the numbers 4.799 and 4.812. 
 
 The salt is decomposed by water, giving, according to the 
 quantity used, either one of the salts containing more mercuric 
 iodide or mercuric iodide itself. It is not dissolved or decom- 
 posed by alcohol. 
 
 CsHgl^.H* Of). This salt is formed only within very nar- 
 row limits from solutions containing a little more mercuric 
 iodide or water than those from which the preceding salt is 
 obtained. These conditions are perhaps most easily reached 
 by dissolving the last salt in a small amount of hot water and 
 cooling. It often happens that the three salts Cs 2 Hg 3 I 8 , 
 
 Pound. 
 
 27.32 27.39 
 
 Calculated for 
 Cs 2 HgI 4 . 
 
 27.31 
 
 21.57 
 
 21.21 
 
 20.53 
 
 51.41 
 
 51.49 
 
 52.16 
 
 100.30 
 
 100.09 
 
 100.00 
 
THE CAESIUM-MERCURIC HALIDES. 229 
 
 CsHglg, and Cs 2 HgI 4 are successively deposited as a solution 
 cools, and it is consequently difficult to obtain the salt under 
 consideration in a pure state, but this was accomplished after 
 a great many trials with varying conditions. The compound 
 forms very thin transparent plates which usually radiate from 
 a point and are often of large size. By pressing on paper they 
 rapidly become opaque. Whether this is caused by molecular 
 rearrangement or loss of water of crystallization is not certain, 
 for, on account of the extreme thinness of the crystals, it was 
 impossible to decide whether a small amount of moisture or a 
 molecule of very unstable water of crystallization was present. 
 Two samples were analyzed. A was air-dried after pressing 
 on paper ; B was quickly dried on paper. 
 
 
 Found. 
 A. 
 
 Calculated for 
 CeHgl,. 
 
 Found. C 
 B. 
 
 Jalculated : 
 
 Caesium. . 
 
 . 18.81 
 
 18.63 
 
 18.25 
 
 18.17 
 
 Mercury . 
 
 . 29.29 
 
 28.01 
 
 28.74 
 
 27.33 
 
 Iodine 
 
 . 51.50 
 
 53.36 
 
 50.98 
 
 52.05 
 
 Water . . 
 
 . . . 
 
 0.00 
 
 2.51* 
 
 2.45 
 
 99.60 100.00 100.48 100.00 
 
 Like all the other iodides, this salt is decomposed by water. 
 
 Cs 2 Hg s l8 is formed under widely different conditions. It is 
 most convenient to prepare it by dissolving Cs 2 HgI 4 in the 
 proper amount of hot water and cooling. It is also formed, 
 in a finely divided condition, by treating the same salt with 
 not too much cold water. The crystals vary considerably in 
 habit, but they can be readily distinguished from the other 
 iodides. A characteristic form is a triangular plate, but plates 
 of different shape and more or less elongated prisms often oc- 
 cur. The following analyses were made of separate crops. 
 Sample C was made by treating Cs 2 HgI 4 with cold water. 
 
 Found. Calculated for 
 
 . A. > B. C. CsjHg 8 I 8 
 
 Cesium . . 13.89 . . . 14.14 14.07 14.13 
 
 Mercury . . 33.76 33.83 31.88 
 
 Iodine . . . 52.07 52.10 52.63 52.96 53.99 
 
 99.72 100.86 100.00 
 * By loss at 100. 
 
230 THE CAESIUM-MERCURIC HALIDES. 
 
 Specific gravity, taken in benzol, 5.14. The salt dissolves 
 in alcohol. It is decomposed by water with the separation of 
 a part of the mercuric iodide. From the solution thus ob- 
 tained, the salts containing less mercuric iodide can be pre- 
 pared by evaporation. 
 
 CsHff 2 T&. When a hot aqueous solution of caesium iodide 
 is saturated with mercuric iodide, this compound is formed on 
 cooling, but, under these conditions, the substance is usually 
 mixed with HgI 2 and often with Cs 2 Hg 3 I 8 . When weak alco- 
 hol is used as a solvent, however, a pure product is obtained 
 without difficulty. It forms slender yellow prisms which be- 
 come red on standing in an aqueous mother-liquor. They are 
 more permanent in the solution when it is alcoholic, but, on 
 drying them by pressing on paper, they quickly assume the 
 red color of mercuric iodide without losing their form. It is 
 probable that the spontaneous decomposition results in the 
 formation of Cs 8 HgaI 8 and HgI 2 . It was necessary to analyze 
 the material which had become red. 
 
 ,, , Calculated for 
 
 Found. 
 
 Csesium .... 11.47 11.39 
 
 Mercury .... 35.73 34.25 
 
 Iodine ..... 52.93 54.36 
 
 100.13 100.00 
 
 The Mixed Double Halides. 
 
 A great deal of labor has been devoted to a study of these 
 compounds in order to find to what extent they could be pre- 
 pared. The results show that caesium chloride and mercuric 
 bromide unite readily, although there is a tendency towards an 
 exchange of halogens and the formation of unmixed salts. It 
 is also noteworthy that, while there is a double chloride as well 
 as a double bromide which is not decomposed by recrystalliza- 
 tion from water, all the chloro-bromides finally yield mercuric 
 bromide when so treated. 
 
 The number of bromo-iodides is less than that of the 
 unmixed salts, for, when attempts are made to prepare com- 
 
THE CESIUM-MERCURIC HALIDES. 231 
 
 pounds containing the larger amounts of mercuric iodide, 
 there is an exchange of halogens and almost pure double 
 iodides are produced. 
 
 Only one compound of mercuric iodide with caesium chlo- 
 ride could be prepared. This is Cs a HgCl 2 I 2 , and the type to 
 which it belongs may probably be considered, on this account, 
 the most stable one of the csesium-mercuric halides. 
 
 It is evident that the mixed salts are not as readily formed 
 as the unmixed, and that the more dissimilar the two halogens 
 are, the less tendency there is to form the mixed compounds.* 
 
 In preparing these salts, containing two different halogens, 
 the halogen of higher atomic weight was always added in 
 combination with the mercury. The methods of preparation 
 are exactly analogous to those by which the unmixed salts 
 are made, so that most of these details will be omitted in 
 describing them. 
 
 The Chloro-bromides. 
 
 In form these all resemble the unmixed salts between which 
 they are intermediate, and all of them are colorless except 
 CsHgClBr 2 , which is pale yellow. 
 
 Calculated for 
 C S8 HgCl,Br 8 . 
 
 Osium .... 48.12 46.10 
 
 Mercury .... 23.80 23.11 
 
 Chlorine .... 16.24 12.30 
 
 Bromine .... 11.82 18.49 
 
 99.98 ioOjW 
 
 The product was made with a very large excess of caesium 
 chloride, and it contained a considerable amount of the double 
 chloride. The analysis corresponds nearly to the formula 
 2Cs 3 HgCl 3 Br 2 + Cs s HgCl 6 . 
 
 O8 2 HgCl z Br 2 . Two products, which were made under 
 different conditions, were analyzed. 
 
 * This point is discussed in connection with the caesium trihalides. (Wells 
 and Penfield, Amer. Jour. Sci., Ill, xliii, 31 and 32.) 
 
232 
 
 THE CESIUM-MERCURIC HALIDES. 
 
 Found. 
 
 Caesium . 
 Mercury . 
 Chlorine . 
 Bromine . 
 
 40.34 
 28.79 
 12.94 
 17.43 
 
 38.86 
 28.58 
 10.48 
 22.07 
 
 99.50 99.99 
 
 Calculated for 
 Cs 2 HgCl 2 Br a . 
 
 38.16 
 28.69 
 10.19 
 22.96 
 100.00 
 
 One of these crops corresponds very closely to the formula, 
 while the other, made in the presence of a greater excess of 
 caesium chloride, contains a little Cs 2 HgCl 4 . 
 
 CsHgClBr z . This has been obtained, like the chloride and 
 bromide, in dimorphous forms. One of these is cubic like the 
 other salts, while the second form crystallizes like the chloride 
 and not like the bromide. The color of both varieties is pale 
 yellow. 
 
 Found. 
 
 
 Cubic Form. 
 
 Orthorhombic Form. 
 Separate Products. 
 
 Calculated for 
 CsHgClBr 2 . 
 
 Caesium . 
 
 , 26.50 
 
 26.97 
 
 26.74 
 
 26.01 
 
 25.17 
 
 Mercury . 
 
 . 38.75 
 
 40.21 
 
 40.05 
 
 38.91 
 
 37.84 
 
 Chlorine . 
 
 . 9.23 
 
 11.32 
 
 11.42 
 
 8.53 
 
 6.72 
 
 Bromine . 
 
 . 25.21 
 
 21.63 
 
 21.94 
 
 26.65 
 
 30.27 
 
 
 99.69 
 
 100.13 
 
 100.15 
 
 100.10 
 
 100.00 
 
 These products evidently contain some of the chloride. 
 The analyses of the first two samples of the orthorhombic salt 
 correspond closely to the formula, 2CsHgClBr 2 + CsHgCl 3 . 
 
 CsHg 2 ClBr. Two separate products, made under different 
 conditions, were analyzed. 
 
 Caesium 
 Mercury 
 Chlorine 
 Bromine 
 
 Found. 
 
 15.48 15.23 
 
 45.72 45.06 
 
 5.75 3.71 
 
 32.30 _36.06 
 
 99.25 100.06 
 
 Calculated for 
 CsHg 2 ClBr 4 . 
 
 14.97 
 
 45.02 
 
 3.99 
 
 36.02 
 
 100.00 
 
 This compound was prepared by recrystal- 
 lizing the preceding salt from water. 
 
THE CESIUM-MERCURIC HALIDES. 233 
 
 f, , Calculated for 
 
 C 8 Hg 5 ClBr 10 . 
 
 Caesium . . . 6.23 6.76 
 
 Mercury . . . 52.77 50.80 
 
 Chlorine . . . 2.85 1.80 
 
 Bromiue . . . 38.19 40.64 
 
 100.04 100.00 
 
 There is a chloride corresponding to this compound, but no 
 bromide was obtained of this type. It forms elongated crys- 
 tals, much smaller than the chloride. The final product, when 
 this salt is recrystallized from water, is mercuric bromide. 
 
 The Bromo-iodides. 
 
 Only three of these compounds have been prepared. When 
 attempts were made to obtain compounds containing larger 
 amounts of mecuric iodide, there was an interchange of halo- 
 gens and nearly pure double iodides were formed. Two such 
 products were analyzed. 
 
 Found. Calculated for Found Calculated for 
 
 Us 2 Ug 3 l 8 . (JsJtlg 2 l 5 . 
 
 Caesium . . . 14.69 14.13 11.62 11.39 
 
 Mercury . . . 33.72 31.88 36.09 34.25 
 
 Bromine . . . 2.19 0.00 2.43 0.00 
 
 Iodine .... 49.58 53.99 49.45 54.36 
 
 100.18 100.00 99.59 100.00 
 
 Cs^HgBr^Iz. This salt resembles the iodide, not the bro- 
 mide, in form. Its color is a pale yellow, intermediate between 
 the brighter iodide and the colorless bromide. 
 
 Caesium . . . 37.21 36.50 
 
 Mercury . . . 19.39 18.30 
 
 Bromine . . . 25.18 21.96 
 
 Iodine .... 18.24 23.24 
 
 100.02 100.00 
 
 . This compound has a very faint tinge of 
 
 yellow. It is apparently dimorphous, although no other salt 
 
234 THE CAESIUM-MERCURIC HALIDES. 
 
 of this type has been made in more than one form. It occurs 
 in very thin plates, like the chloride, bromide, and chloro- 
 bromide, and in stout monoclinic crystals like the iodide. 
 The limits of the conditions under which the plates are made 
 are very narrow, and it is difficult to obtain them free from 
 the dimorphous crystals. As the solution cools, however, the 
 plates are deposited first, and, with the proper dilution, it is 
 possible to remove them and get the mother-liquor pressed out 
 with paper before the other crystals begin to form. There 
 is no difficulty in preparing the other modification of the 
 compound. 
 
 Found. 
 
 m,. T> lofoo Orthorhombic Calculated for 
 Thin Plates. Crystals. Cs.,HgBr 2 I 2 . 
 
 Cesium . . . 30.71 30.20 30.23 
 
 Mercury . . . 24.14 23.86 22.73 
 
 Bromine . . . 21.05 17.91 18.18 
 
 Iodine. . . . 24.23 28.50 28.86 
 
 100.13 100.47 100.00 
 
 It is noticable that the plates, which resemble the bromide 
 in form, contain a small excess of bromine and a correspond- 
 ing deficiency of iodine. 
 
 CsHgBrli. Only one form of this compound has been 
 prepared, although three other salts of this type are dimor- 
 phous. Its form is monoclinic, like one modification of the 
 bromide, and it is pale yellow in color. 
 
 ,, , Calculated for 
 
 Found. CsHgBrI 2 . 
 
 Caesium . . . 20.26 19.94 
 
 Mercury . . . 31.44 29.99 
 
 Bromine . . . 13.35 11.99 
 
 Iodine .... 34.39 38.08 
 
 99.44 100.00 
 
 The Chloro-iodide, Cs z Hg 01 2 I 2 . 
 
 This is the only combination of caesium chloride and mer- 
 curic iodide that could be produced. It is formed only in 
 very concentrated solutions containing a great excess of 
 
THE CAESIUM-MERCURIC HALIDES. 235 
 
 caesium chloride. Its form is different from any other salt of 
 the type, for it occurs in slender, radiating needles. It is 
 snow-white in color, and when it is brought in contact with 
 water it instantly becomes bright red from the formation of 
 mercuric iodide. Two entirely separate crops were analyzed. 
 
 Found. 
 
 Caesium . . . 33.38 33.14 33.63 
 
 Mercury . . . 26.71 . . . 25.28 
 
 Chlorine . . . 8.87 9.01 8.98 
 
 Iodine. . . . 30.85 30.17 32.11 
 
 99.81 100.00 
 
 When it was attempted to make a chloro-iodide containing 
 more mercuric iodide than this, a nearly pure double-iodide 
 was formed by exchange of halogens. 
 
 Calculated for 
 
 Csesium . . . 13.76 
 Mercury . . . 33.49 
 Chlorine . . . 0.16 
 Iodine .... 50.74 
 98.15 
 
 The investigation of double halides will be continued hi this 
 laboratory, and it is hoped that a further study of the caesium 
 salts will lead to a better knowledge of this class of compounds 
 in general than we now possess. 
 
 In conclusion, it gives me pleasure to express my gratitude 
 to my colleague, Professor Penfield, for his hearty co-operation 
 in undertaking the crystallographic examination of the com- 
 pounds which have been described. His results have been 
 freely used in the foregoing descriptions, and they will be 
 given in detail in a future article. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 May, 1892. 
 
ON THE CRYSTALLOGRAPHY OF THE CESIUM- 
 MERCURIC HALIDES.* 
 
 BY S. L. PENFIELD. 
 
 THE salts to be described in this paper were prepared by 
 Prof. H. L. Wells, and their chemical description has been 
 given by him in the preceding paper. 
 
 The crystals were all measured on a Fuess reflecting goni- 
 ometer, model II, and great pains were taken to select the best 
 measurements as fundamental. In a few cases, where the 
 crystals were very small and the reflections of the signal, 
 therefore, rather broad, the mean of a series of measurements 
 was used. The axial ratios are given in tabular form at the 
 beginning of each separate chemical type, and the fundamental 
 angles, from which these are derived, are marked by an asterisk 
 in the table of angles accompanying each salt. 
 
 Type 3 : 1. 
 
 & : b : c 
 0.7976 : 1 : 0.6605 
 0.7882 : 1 : 0.6527 
 0.7966 : 1 : 0.6656 
 0.5362 : 1 : 0.97975 
 0.8043:1:0.6532 
 Approx. like Cs 3 HgI 5 
 
 The first three salts have exactly 
 the same habit and crystallize in 
 slender prisms, attached at one 
 end and terminated at the other 
 by faces which are arranged with 
 monoclinic symmetry, Figs. 1, 2, 
 and 3. The crystals were seldom 
 over 1 mm. in diameter, but the 
 faces were perfect and admitted of 
 
 Cs 8 HgCl 6 
 Cs 3 HgCl 3 Br 2 
 Cs 8 HgBr 6 
 Cs 8 HgI 6 
 
 Cs 3 HgBr 8 I 2 
 1. 
 
 Orthorhombic, monoclinic hemihedrism 
 Orthorhombic, monoclinic hemihedrism 
 Orthorhombic, monoclinic hemihedrism 
 Orthorhombic, sphenoidal hemihedrism 
 or f a : b : c = 
 Orthorhombic, sphenoidal hemihedrism 
 
 m 
 
 m 
 
 771 
 
 accurate measurement. The forms and angles are : 
 
 * Amer. Jour. Sci., xliv, October, 1892. 
 
CESIUM-MERCURIC HALIDES. 237 
 
 m, 110, 1 d>, OT1, '1-T jp, 111, 1 
 
 d, on, i-r 6, 021, 2-r y, ITI, a 
 
 Measured. Calculated. Measured. cWiilated. Measured! 3 Calculated. 
 
 m A m, 110 A 110 = *77 9' ... 76 33' 76 30' *77 5' ... 
 
 m A/), HO A 111 = *43 21' ... *43 29' ... *43 6' ... 
 
 m AP, 110 A 111= 80 41 X 8041i' 80 37' 80 36$' 
 
 m A d, 110 A Oil = 69 55' 69 54' 70 14' 70 13' 69 49' 69 48' 
 
 m A e, 110 A 021 = 60 8' 60 11' 60 10' 60 7' 
 
 d A e, Oil A 021 = 19 29' 19 26' 19 28' 19 25' 
 
 d A d',011 A Oil = 66 53' 66 63$' *66 16' ... 67 16' 67 18' 
 
 d A/>, Oil A 111 = 34 39' 34 39' 34 37' 34 44' 34 47' 34 4^ 
 
 The crystals have orthorhombic optical properties. When 
 lying on their prismatic faces all show in polarized light an ex- 
 tinction parallel to the vertical axis, and in convergent light a 
 trace of the ring system can be seen, indicating that the plane 
 of the optical axes is the base. 
 
 We have here an excellent illustration of monoclinic hemi- 
 hedrism in the orthorhombic system. Among all of the crystals 
 which were examined, there was not one which had a holo- 
 hedral termination. The forms d' and p 1 , when present, were 
 always smaller than the corresponding forms d and jt?, while e 
 was only observed to the right above. Also the right-handed 
 vertical edge of the prism showed a tendency toward a skele- 
 ton-like growth, which was not observed to the left. In 
 measuring the crystals great pains were taken to detect a 
 monoclinic character by the angles, but none could be found. 
 Of course the three salts may be regarded as monoclinic, 
 with an angle /3, differing so little from 90 that it cannot 
 be detected by the goniometer ; but against such a supposition 
 are the arguments that the crystals have orthorhombic optical 
 properties, and while there is a variation in the axial ratios of 
 the series as bromine is substituted for chlorine, there is no 
 change in the angle & as would be expected if the salts were 
 monoclinic. In this connection it is interesting to note that 
 while the chloride and bromide are very similar in their axial 
 ratios, the chemically intermediate chloro-bromide is not 
 crystallographically intermediate. 
 
238 
 
 CRYSTALLOGRAPHY OF THE 
 
 At the present time there seems to be no other known 
 compound which illustrates this hemihedrism. Different sub- 
 stances, which have been referred to this class, as datolite 
 or wolframite, for example, have been shown by accurate 
 measurement, or a study of their optical properties, to be truly 
 monoclinic. Prof. P. Groth, in the last edition of his " Phy- 
 sikalische Krystallographie," has not mentioned this hemi- 
 hedrism as a possibility in the orthorhombic system, although 
 in the former edition of his work and in most treatises on 
 crystallography it is recognized. 
 
 The different crops of Cs 3 HgI 6 which were examined showed 
 a great variety in habit, represented by Figs. 4-8. The hemi- 
 hedral development is not always strongly marked, and forms 
 like Figs. 5 and 6 are the commonest. The crystals sometimes 
 measured over 5 mm. in diameter and gave excellent reflections. 
 
 Only one crop of Cs 3 HgBr 3 I 2 was examined. The crystals 
 were in the form of sphenoids, Fig. 9, some of them over 10 mm. 
 in diameter, but the faces were curved and striated and only 
 approximate measurements could be made. 
 
 The forms which were observed are : 
 b, 010, i* d, 102, H s, 021, 2-r 
 
 p 
 
 c, 001, r, Oil, 1-T p, 111, 1 
 
 The angles of Cs 8 HgI 5 are : 
 
 112, 
 
 Measured. Calculated. 
 
 c A p, 001 A 111 = *64 15' 
 
 p A ;/, 111 A 111 = *50 23' 
 c A r, 001 A Oil = 44 23' 
 c A s, 001 A 021 = 62 58' 
 
 44 25' 
 62 58' 
 
 Measured. 
 
 c A o, 001 A 112 = 46 V 
 o A o', 112 A 112 = 39 65' 
 o A o', 112 A 112 = 87 57' 
 o A d, 112 A 102 = 19 50' 
 p A e, 111 A 121 = 17 48' 
 
 121,-2-2 
 
 Calculated. 
 46 2' 
 39 46' 
 87 56' 
 19 53 X 
 18 4 X 
 
CAESIUM-MERCURIC HALIDES. 239 
 
 Both Cs 8 HgI 6 and Cs 8 HgBr 8 I 2 cleave distinctly, parallel to 
 the base, but the crystals are very brittle and usually break 
 with a conchoidal fracture. Crystals of the former, which are 
 tabular parallel to the base, show in convergent polarized light 
 a bisectrix normal to c, 001 ; the plane of the optical axes is 
 the macropinacoid, and their divergence is large. 
 
 Type 2 : 1. 
 
 a : b : c 
 
 Cs 2 HgCl 4 Not measured. 
 
 Cs 2 HgBr 4 Orthorhombic 0.6706 : 1 : 1.4715 
 
 Cs 2 HgCl 2 Br 2 Orthorhombic 0.567 : 1 : ... 
 
 Cs 2 HgCl 2 I 2 Orthorhombic Not measured. 
 
 Cs 2 HgI 4 Monoclinic 1.3155:1:0.9260 = 69 56' 
 
 Cs 2 HgBr 2 I 2 Monoclinic Approximately like Cs 2 HgI 4 . 
 
 The crystals of Cs 2 HgCl 4 were too thin to measure. 
 
 Both Cs 2 HgBr 4 and Cs 2 HgCl 2 Br 2 crystallize in 
 thin rectangular plates; those of the former were 
 sometimes several centimeters long, but seldom over 
 J mm. thick, and had the habit shown in Fig. 10. 
 The crystals of the latter salt were very much 
 thinner. The plates were often grouped, with the 
 large pinacoid faces slightly divergent, and isolated 
 crystals, suitable for exact measurement, were only 
 occasionally found. 
 
 The forms which were observed on Cs 2 HgBr 4 are : 
 
 ft, 010, f-r c, 001, m, 110, / d, Oil, 1-T j>, 221, 2 
 
 Kg. 10. 
 
 and on Cs 2 HgCl 2 Br 2 , 6, m and a second prism 130, i-3. The 
 end faces could not be made out. 
 
 The angles of Cs 2 HgBr 4 are : 
 
 Measured. Measured. Calculated. 
 
 m A m, 110 A 1TO = *59 25' m A Pt 110 A 221 = 33 58 J' 33 58' 
 
 6 A 
 
 On this salt the dome d is always small and frequently 
 wanting. The pyramid p was only observed on a few crystals. 
 In convergent polarized light a bisectrix may be seen normal 
 
240 CRYSTALLOGRAPHY OF THE 
 
 to 5, 010. The plane of the optical axes is the macropinacoid, 
 and their divergence is so large that they cannot be measured 
 in air, but in <2-monobromnaphtaline the following values were 
 obtained : 
 
 2H = 80 12' for yellow, Na flame. 
 2H = 85 23' for red, Li flame. 
 
 The dispersion is strong p>v. The acute bisectrix is axis 
 of least elasticity, the double refraction is therefore positive. 
 The only angles on Cs 2 HgCl 2 Br 2 which were measured are : 
 
 m A m, 110 A 110 = *59 6' and 110 A 130 = approx. 30 53', calculated 30 59' 
 
 In convergent polarized light a bisectrix may be seen normal 
 to 5, 010. The plane of the optical axes is the macropinacoid, 
 and their divergence is large. The axis of greatest elasticity 
 is normal to b. 
 
 Only very fine needles of Cs 2 HgCl 2 I 2 were obtained, which 
 were too small for measurement. These appeared under the 
 microscope as striated prisms, with their obtuse edges rounded 
 by oscillatory combinations. In polarized light they show a 
 parallel extinction, and in convergent light a biaxial interfer- 
 ence figure, the plane of the optical axes being the vertical 
 pinacoid. The acute bisectrix is axis of least elasticity. 
 
 The crystals of Cs 2 HgI 4 , which were frequently several cen- 
 timeters in diameter, showed a variety of habits represented 
 
 13. 
 
 42 
 
 in Figs. 11, 12 and 13. The crystals of the latter habit are 
 usually attached at one end and taper toward the free ex- 
 tremity, owing to a tendency to develop vicinal pyramids in 
 the zone d-b. 
 
CAESIUM-MERCURIC HALIDES. 
 
 241 
 
 The crystals of Cs 2 HgBr 2 T 2 , which were 
 examined, were about 2 mm. in diameter and 
 had the simple habit shown in Fig. 14. The 
 faces were rounded and uneven, so that only 
 approximate measurements could be made. 
 
 The forms and angles are : 
 
 a, 100, i-l 
 
 b, 010, i- 1 
 
 c, 001, 
 m, 110, / 
 
 Cs 2 HgI 4 . 
 Measured. Calculated. 
 
 c A a, 001, 100 = *69 56' 
 
 m A m, 110 A T10 = *77 58' 
 
 a A d, TOO A 201 = *41 25' 
 
 c A e , 001 A Oil = 40 55' 
 
 d, 201, 2-i 
 
 e, Oil, 14 
 
 Cs,HgBr 2 I 2 . 
 Measured approximately. 
 
 66 41' to 66 47' 
 77 16' to 77 57' 
 
 41 V 
 
 The cleavage of both salts is perfect parallel to the base, 
 less so parallel to the clinopinacoid. With Cs 2 HgI 4 the plane 
 of the optical axes is at right angles to the symmetry plane, 
 and clinopinacoid cleavage sections show in convergent polar- 
 ized light an obtuse bisectrix, which is axis of least elasticity. 
 The axis of greatest elasticity makes an angle of about 50 
 with the vertical axis in the acute angle j3. 
 
 Type 1 : 1. 
 
 a : b : c 
 
 CsHgClg Isometric and Orthorhombic 0.57735 : 1 : 0.40884 
 CsHgClBr 2 Isometric and Orthorhombic Approximately like the above. 
 CsHgBrg Isometric and Monoclinic 1.0124 : 1 : 0.70715 = 87 7' 
 CsHgBrLj Monoclinic 0.978 : 1 : 0.743 = 87 3' 
 
 CsHglg Not measured. 
 
 The first three compounds are dimorphous, and, from solu- 
 tions containing an excess of alkali halide, they all crystallize 
 in cubes. These sometimes have their edges truncated by 
 small dodecahedron faces, less often bevelled by 210, i-2. The 
 crystals show a slight action on polarized light and give an 
 extinction parallel to the diagonals of the cube, but this 
 anomaly is probably due to some internal tension, for when 
 crushed the fragments are isotropic. No cleavage could be 
 detected. 
 
 16 
 
242 CRYSTALLOGRAPHY OF THE 
 
 CsHgCls was repeatedly recrystallized from water and 
 always two types were observed. One of these was confined 
 
 to those crystals which 
 
 15. 16- were attached to the sides 
 
 of the beaker, while those 
 which grew more in the 
 interior had an entirely 
 different habit. The 
 
 crystals of the first type averaged about 2 mm. in greatest 
 diameter and had the habit shown in Figs. 15 and 16. The 
 forms and angles are as follows : 
 
 a, 100, i-1 m, 110, / e, 101, 1-* d, 021, 2-i 
 
 b, 010, i-i n, 130, i-3 /, 201, 2-i p, 111, 1 
 
 Measured. Calculated. Measured. Calculated. 
 
 m A ro, 110 A HO = *60 0' ... a A p, 100 A 111 = 56 45' 66 45$' 
 
 d A d, 021 A 021 = *101 27f ... m A />, 110 A 111 - 71 33$' 71 33' 
 
 P Ap, 111 A 111 = 66 28$' 66 29' p A d, 111 A 021 = 36 54' 36 54' 
 
 P Ap, 111 A 111 = 36 54' 36 54' a A e, 100 A 101 = 54 42$' 54 42' 
 
 m A p, 110 A 111= 60 43$' 5043f a A/, 100 A 201 = 35 13' 86 13$' 
 
 The crystals were brilliant and gave wonderful reflections. 
 The prismatic angle was measured repeatedly and found to be 
 60, and the forms could be referred to the hexagonal system, 
 making the m and b faces a prism of the first order, a and n a 
 prism of the second order, and p and d the unit pyramid. 
 There was nothing, however, in the development of the faces 
 to suggest hexagonal symmetry. Thin sections were pre- 
 pared, hoping that the optical properties would throw some 
 light upon the form, but they showed only a very weak double 
 refraction, in fact they appeared almost like isotropic sections, 
 so that no satisfactory conclusions could be drawn. 
 
 The crystals of the second type were spearhead-shaped, 
 Fig. 17, and grew out into the centre of the solution, either 
 attached to one another by the acute solid angles, or to a 
 slender, parallel growth of crystals, which served as a sort of 
 stem. The crystals which are about 5 mm. in length are com- 
 plicated and perplexing, and the faces are developed with tri- 
 
CAESIUM-MERCURIC HALIDES. 
 
 243 
 
 clinic symmetry, although they can be referred to the axes of 
 the first type. The most prominent faces are shown in the 
 figure, while the distribution of all those which gave distinct 
 reflections are given in the spherical projection, Fig. 18. The 
 forms which were observed are given as if they belonged to a 
 triclinic crystal and are : 
 
 Fig. 17. 
 
 b, 010, i-i 
 m', 1TO, '/ 
 /, 201, '2-? 
 d, 021, 2-T 
 
 Fig. 18. 
 
 JP, HI, 1' 
 /, Til, ,1 
 
 ?", ni, i, 
 
 2, 132, f-S 
 
 Fig. 19. 
 
 Fig. 20. 
 
 r"', 132, 'f-3 y w , 7f5,'|-f 
 
 /, T31, ,3-3 z", T^7, f 9' 
 
 w, T91, ,9-9 v, 1131, 11-V-' 
 af, T151, ,15-15 
 
 The crystals gave excellent reflections, and only occasionally 
 a slight striation interfered with making accurate measure- 
 ments. All of the forms were observed on two crystals, and 
 probably others could have been found by measuring a larger 
 number. 
 
 The crystals of CsHgClBr 2 have a similar habit, Fig. 19, 
 and the distribution of all of the faces which gave distinct re- 
 flections is given in the spherical projection, Fig. 20. This 
 salt is more insoluble than the chloride, and the crystals are 
 consequently much smaller, not over 1 J mm. in greatest diam- 
 eter. All of the forms given above for the chloride were 
 observed except z and v, and in addition : 
 
 d', 051, '2-r /", 1T1, '1 /', T31, 3-3, u', T71, 7-7 t, 2T27, 'y-6 
 
 The crystals gave very good reflections, considering their 
 size, and the best measurements agreed so well with those of 
 
244 CRYSTALLOGRAPHY OF THE 
 
 the cliloride that no attempt was made to calculate a new axial 
 ratio. The most marked difference in the two salts is the 
 development of the zone p" z" q'" y'" in the chloride and r n 
 d' t'" q f " p" f f'w the chlorobromide. 
 
 The measured and calculated angles are as follows : 
 
 CsHgCl 3 . CsHgClBr 2 . Calculated. 
 
 m' A /, 1TO A 201 44 45' 45 6' 44 
 
 / A p, 201 A 111 26 35' 26 27' 26 
 
 p A q, 111 A 132 18 27' 18 34' 18 27' 
 
 q A d, 132 A 021 18 26' 18 23' 18 27' 
 
 d A r, 021 A T31 26 44' 26 26' 26 34J' 
 
 x A r, T151 A T31 33 40' 33 20' 33 40^' 
 
 v A r, T91 A T31 26 36' 26 27' 26 33' 
 
 u A r, 171 A T31 ... 21 35' 21 47$' 
 
 r A /, T31 A Til 26 33' 26 37' 26 34^' 
 
 p' A /', Til A TT1 36 52$' 36 37' 36 54' 
 
 b A r", OTO A 151 44 44' ... 44 58' 
 
 y m A ?'", 7S5 A 132 22 13' ... 22 13' 
 
 z" A /', T37 A TT1 28 38' ... 28 35' 
 
 q'" A p", 132 A TT1 50 54' ... 50 49' 
 
 y"' A q, 735 A 132 62 29^' 62 22' 62 28' 
 
 .p" A v, III A TT3T 60 46' ... 60 45' 
 
 / A v, T11 A TT3T 66 53' . . . 66 59' 
 
 / A /", 201 A 1T1 ... 26 38' 26 34$' 
 
 /"A q'", ITlA 132 ... 18 23' 18 27' 
 
 q'" A , 132 A 021 ... 18 34' 18 27' 
 
 t'" A d', 2T27 A 021 ... 10 13' 10 19' 
 
 q A w, 132 A T91 50 59' 51 4' 50 46' 
 
 q A /", 132 A 1T1 ... 50 50' 50 49' 
 
 q A ?'", 132 A 132 60 3' ... 60 4' 
 
 It will be seen from the spherical projection that the forms 
 of CsHgClBr 2 lie mostly in three zones, suggestive of hexag- 
 onal rhombohedral symmetry, although there is nothing in 
 the arrangement of the faces, and still less with CsHgCl s , to 
 indicate that this is correct. The crystals of CsHgClBr 2 have 
 a slightly stronger action on polarized light than those of 
 CsHgClg. When lying on the large q faces, both show an 
 extinction parallel to the edges between p, q and d. 
 
CESIUM-MERCURIC HALIDES. 
 
 245 
 
 On a crystal of CsIIgCl 3 the faces in the zone p, q, d, r 
 made prisms, which served for the determination of the follow- 
 ing indices of refraction: 
 
 Prism of 36 54', 021 A III, n, y , Na flame =: 1.791 n, r , Li flame = 1.779 
 Prism of 63 28^, 131 A 111, n, y , " " =1.792 n, r , " " =1.779 
 
 The crystal was of course very small and the refracted rays 
 were not very bright, but the latter were well defined and the 
 double refraction was not strong enough to separate them into 
 two distinct rays. 
 
 The author cannot give any satisfactory explanation of 
 these curious forms. They seem to illustrate a tetartohedral 
 development of the faces of an orthorhombic crystal, resulting 
 in a figure with triclinic symmetry. The mathematical rela- 
 tions have been very carefully determined and the facts given. 
 It is hoped that a further study will throw some light on the 
 subject. 
 
 The different crops of CsHgBr 3 which were examined 
 showed a variety of habits, represented by Figs. 21, 22, and 23. 
 
 2L 
 
 22. 
 
 23. 
 
 The forms and angles are as follows : 
 
 c, 001, 
 , 110, / 
 
 e, 201, 2-1 
 d, T01, \-l 
 
 Measured. Calculated. 
 
 o, Til, 
 p, 221, 
 
 1 ?/, 261, 6-3 
 
 2 x, T31, 3-3 
 
 Measured. Calculated. 
 
 c A d, 001 A 101 = *35 6^ 
 
 c A e, 001 A 201 = 52 28J' 52 30 
 
 c A o, 001 A 111 = 45 40 ; 45 49' 
 
 m Am, 
 
 d A x, 101 A 131 = 60 41' 60 40' 
 
 e A p, 201 A 221 = 38 46' 38 46 \' 
 
 e A y, 201 A 261 = 67 21' 67 28' 
 
 The crystals are seldom over 5 mm. in diameter and some- 
 times have a hemimorphic development, although this is not 
 
246 
 
 CRYSTALLOGRAPHY OF THE 
 
 always apparent. In the prevailing type, Fig. 22, there is 
 perhaps a tendency for 221 to predominate over 221, but this 
 is not great. The pyramids x and y were 
 24 - observed only with hemimorphic develop- 
 
 ment, Fig. 23. The crystals were tested 
 for pyro-electricity, but no satisfactory re- 
 sults were obtained, which is perhaps 
 owing to their small size. The crystals of CsHgBrI 2 were 
 about 2 mm. in diameter and had the habit shown in Fig. 24, 
 which is quite different from that of the bromide. The forms 
 and angles are as follows : 
 
 , 010, i-i c, 001, I, 320, i-f 5, 034, f 4 
 
 Measured. Calculated. 
 
 I A s, 320 A 034 = *72 22' ... 
 I A s, 320 A 034 = 76 16' 76 50' 
 
 Measured. 
 
 I A I, 320 A 320 = *66 8' 
 b A s, 010 A 034 = *60 54' 
 
 The basal planes were curved and uneven, so that no satis- 
 factory measurements could be made from them, and the other 
 faces, although bright, did not give very satisfactory reflec- 
 tions. The crystals show in convergent polarized light an 
 optical axis, almost normal to the base, the plane of the 
 optical axes being the clinopinacoid. 
 
 Type 2 : 3. 
 
 Cs 2 Hg 3 I 8 Monoclinic, hemihedral 
 
 a : b : c 
 0.3438 : 1 : 0.3544 
 
 = 71 55J" 
 
 26. 
 
 The crystals of this salt have a curious development. 
 Some of the most conspicuous forms are triangular plates, 
 Fig. 25, while Fig. 26 is a projection of the same upon the 
 
CESIUM-MERCURIC HALIDES. 247 
 
 clinopinacoid. These crystals are terminated above by a basal 
 plane and below by pyramidal faces, which gives a curious 
 hemimorphic development in the direction of the symmetry 
 plane. A variety of habits was observed, long prismatic, 
 skeleton forms and simple shapes like Fig. 27, but in almost 
 all of these the hemihedral character was prominent. The 
 crystals frequently measured over 10 mm. in greatest diameter. 
 The faces were bright and gave excellent reflections. The 
 forms and angles are as follows : 
 
 b, 010, i-i c, 001, m, 110, / p, 111, 1 
 
 The pyramid was observed only with hemihedral develop- 
 ment. 
 
 Measured. Measured. Calculated. 
 
 CAW, 001 A 110 = *72 51' m *p, 110 A 11T = *50 26' ... 
 m A m, 110 A 1TO = *36 12' b A p, 010 A 111 = 74 17' 74 14' 
 
 Two cleavages were observed, one perfect parallel to the 
 clinopinacoid, a second less perfect parallel to the base. In 
 polarized light clinopinacoid tables give an extinction, inclined 
 about 23 to the vertical axis in the acute angle 0. Basal 
 plates show in convergent light an optical axis not far removed 
 from the centre of the field. The plane of the optical axes is 
 the clinopinacoid. 
 
 These crystals furnish an excellent illustration of inclined 
 faced hemihedrism, as recently developed by Prof. Geo. H. 
 Williams,* who has shown that it is of frequent occurrence 
 
 on pyroxene. 
 
 Type 1 : 2. 
 
 >, : b : c 
 
 B = 78 64' 
 
 
 
 a : b : c 
 
 CsHg 2 Cl 6 
 
 Monoclinic 
 
 1.6099 : 1 : 1.3289 
 
 CsHg 2 ClBr 4 
 
 Orthorhombic 
 
 0.586 : 1 : ... 
 
 CsHg 2 Br 6 
 
 Orthorhombic 
 
 0.590 : 1:1.15 
 
 CsHg 2 I 6 
 
 Not measured. 
 
 
 s was 
 made in slender / 
 
 lath-shaped crys- * * 28 
 
 tals, over 10 mm. 
 
 * Amer. Jour. Sci., xxxviii, 115, 1889. 
 
248 
 
 CRYSTALLOGRAPHY OF THE 
 
 long in the direction of the symmetry axis, but not over J mm. 
 in diameter. Fig. 28 represents a simple, and 29 a twin 
 crystal, with the orthopinacoid as twinning plane. The forms 
 and angles are as follows : 
 
 a, 100, i-l 
 
 b, 010, i-i 
 
 c, 001. 
 m, 110, I 
 
 d, Oil, 14 
 
 Two orthodomes were also identified, 101 and 201, but they 
 were very small and yielded only approximate measurements. 
 
 Measured. 
 
 a A c, 100 A 001 = *78 54' CAW, 001 A 
 a A m, 100 A 110 = *57 40' m A p, 110 A 111 = 31 12' 
 6- A d, 001 A Oil = *52 31' a A p, 100 A 111 = 58 4' 
 
 b A p, 010 A 111 = 47 19' 
 
 Measured. Calculated. 
 
 = 84 5' 84 5' 
 31 8' 
 58 3' 
 47 19' 
 
 The plane of the optical axes is at right angles to the sym- 
 metry plane, and the obtuse bisectrix is nearly nor- 
 mal to the base. 
 
 Both CsHg 2 ClBr 4 and CsHg 2 Br 5 were made in 
 rectangular tablets, Fig. 30, which were not over 
 1J mm. in greatest diameter and were veiy thin. 
 Twins were common, with the unit prism as twin- 
 ning plane, and the plates often penetrated at angles 
 of about 60 and 120, reminding one of little 
 cerussite twins. 
 
 The forms and angles are as follows : 
 
 CsHg 2 ClBr 4 . 
 
 b, 010, i-i 
 
 , 110, / 
 
 d, Oil, 1-t 
 
 CsHg 2 Br 5 . 
 
 b, 010, Vt 
 n, 120, i-2 
 
 d, Oil, i-i 
 
 e, 014, J-T 
 
 Measured. Calculated. Measured. Calculated. 
 
 m Am, 110 A 110 = *eO 44' ... b A 6, twin = *61 5' ... 
 b A b, twin = 60 35' 60 44' 6 A rf, 010 A Oil = *41 0' ... 
 
 6 A n, 010 A 120 = 40 20' 40 17' 
 
 b A e, 010 A 014 = *73 25' 73 57' 
 
CESIUM-MERCURIC HALIDES. 
 
 249 
 
 31. 
 
 Type 1:5. 
 
 a : b : c 
 
 CsHg 5 Cl n Monoclinic 0.7233 : 1 : 0.4675 = 85 51' 40" 
 CsIig 5 ClBr 10 Monoclinic 0.7111 : 1 : 0.4561 )8 = 85 29' 
 
 The chloride was made in prismatic crystals, 
 fully 10 mm. long, and having the habit shown in 
 Fig. 31. The forms and angles are as follows : 
 
 m, 110, / d, Oil, 14 e, TOl, 1-e /, 101, - 1-1 
 The dome / was usually wanting. 
 
 Measured. Calculated. Measured. Calculated, 
 
 m A m, 110 A 110 = *71 37' ... m A d, 110 A Oil = 72 31' 72 31' 
 
 d A d, Oil A Oil = *50 0' ... m A d, 110 A Oil = 78 49' 78 48' 
 
 m A e, 110 A 101 = *66 8' ... d A /, Oil A 101 = 39 30' 39 29|' 
 
 d A e, Oil A TOl = 41 21' 41 20' e A /, 101 A 101 = 65 43' 65 42' 
 
 The chlorobromide CsHg 5 ClBri is much 
 more insoluble than the chloride and was 
 made in crystals, which were not over 
 mm. in greatest diameter. The habit is 
 shown in Fig. 32, and is very different 
 from that of the chloride. The forms and angles are as 
 follows : 
 
 m, 110, / d, Oil, 14 e, TOl, l-i 
 
 Measured. Measured. Calculated. 
 
 m A m, 110 A 1TO = *70 40' d A , Oil A 101 = *40 58' ... 
 d A d, Oil A OT1 = *48 54' m A d, 110 A Oil = 72 40' 72 40' 
 
 The crystals are strongly double refracting, and the little 
 tables show in convergent polarized light a bisectrix nearly 
 normal to e. The plane of the optical axes is the clinopina- 
 coid and the optical axial angle is small. The inference phe- 
 nomena are very interesting when observed through colored 
 glasses. In the hyperbola position the figure is almost uni- 
 axial when viewed through red glass, while with blue the 
 hyperbolae are separated, probably as much as 15-20. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 June, 1891. 
 
ON THE CESIUM- AND THE POTASSIUM-LEAD 
 
 HALIDES* 
 
 BY H. L. WELLS. 
 
 As a continuation of the work on double halides, in this 
 laboratory,! a study of the caesium-lead salts has been under- 
 taken by Messrs. G. F. Campbell, P. T. Walden, and A. P. 
 Wheeler. These gentlemen have carried out the investigation 
 with much enthusiasm and skill, and I take pleasure in ex- 
 pressing my obligations to them. They have established the 
 following salts : 
 
 Cs 4 PbCl 6 Cs 4 PbBr 6 
 
 CsPbCl 3 CsPbBr 8 t CsPbI 8 
 
 CsPb 2 Cl 6 CsPb 2 Br 5 
 
 These results showed the existence of three types of lead 
 double halides, the first of which fails to conform with Rem- 
 sen's law concerning the composition of this class of bodies. 
 
 Since the recent investigations of Remsen and Herty || had 
 indicated the existence of only a single type of potassium-lead 
 halides, a new investigation of these seemed desirable, espe- 
 cially since these authors had denied the existence of Boullay's 
 salt, If K 4 PbI 6 , which corresponds to one type of the new 
 caesium compounds. I have, therefore, undertaken this work, 
 and, as a result, have obtained the following salts : 
 
 ....... K 2 PbBr 4 .H 2 ....... 
 
 3KPbCl 8 .H 2 :LBr..H, KPbI 8 ,2H 8 
 
 KPb 2 Cl 6 KPb 2 Br 6 
 
 * Amer. Jour. Sci., xlv, February, 1893. 
 t Ibid., Ill, xliv, 155, 157, and 221. 
 $ This compound is dimorphous. 
 
 Amer. Chem. Jour., xi, 296. || Ibid., xiv, 107. 
 
 , T Ann. Chim. Phys., II, xxxiv, 336 (1827). 
 
CAESIUM- AND POTASSIUM-LEAD HAL1DES. 251 
 
 It is to be noticed that neither Boullay's iodide nor any 
 corresponding chloride or bromide was obtained among these 
 salts. On the other hand, the compound K 2 PbBr 4 .H 2 O be- 
 longs to a type which had not been discovered among the 
 caesium salts, so that, taking the caesium and potassium series 
 together, the existence of four types of double lead halides is 
 shown. 
 
 The compound K 2 PbBr 4 , the anhydrous form of the salt 
 just mentioned, is ascribed to Lb'wig,* but although iodides 
 belonging to the same type have been described, K 2 PbI 4 .4H 2 O 
 by Ditte f and K 2 PbI 4 .2H 2 O by Berthelot, J neither Remsen 
 and Herty nor I have been able to prepare them. Although 
 these iodides and Boullay's salt, K 4 PbI 6 , belong to types which 
 certainly exist, I am inclined to believe, with Remsen and 
 Herty, that the products which gave these formulae were mix- 
 tures of KPbI 8 .2H 2 O and KL The absence of more than one 
 iodide in the caesium series strengthens this view. 
 
 Remsen and Herty obtained the salt KPbI 8 .2H 2 O under 
 wide variations of conditions, and I have confirmed their 
 results. This salt was first obtained by Boullay and -ana- 
 lyzed by him, after drying over lime, in an anhydrous condi- 
 tion. Berthelot || has described a compound, K 4 Pb 3 I 10 .6H 2 O, 
 which differs but slightly in required composition from the 
 above salt, and his description of it agrees with that com- 
 pound. There is no doubt, therefore, that he really obtained 
 the compound KPbI 8 .2H 2 O and that his analyzed products 
 were slightly contaminated with potassium iodide. Berthelot 
 attributes K 4 Pb 3 I 10 to Boullay. The latter chemist, however, 
 derived the correct formula, equivalent to KPbI 8 , from his 
 analysis, but since this did not agree closely with theory, 
 Gmelin^f derived the above-mentioned formula from it, and 
 this has been frequently copied in more recent chemical 
 literature. 
 
 * Gmelin's Handbook, English ed. of 1850, v. 162. 
 t Ann. Chim. Phys., V, xxiv, 226, 1881. J Ibid., xxix, 289, 1883. 
 Ibid., II, xxxiv, 336, 1827. || Ibid., V, xxix, 289, 1883. 
 
 T Handbook, English ed., 1850, v, 161. 
 
252 ON THE CsESIUM- AND THE 
 
 Schreinemakers,* in connection with an investigation on 
 the equilibrium of the double salt of iodide of lead and potas- 
 sium in aqueous solution, has assumed that Ditte's formula 
 was correct as far as the composition of the anhydrous compound 
 was concerned. By making a number of water determina- 
 tions, without determining lead, potassium, or iodine, he 
 arrived at the formula K 2 PbI 4 .2iH 2 O. It is absolutely cer- 
 tain, from his description of the salt and his method of pre- 
 paring it, that he had the compound KPbI 3 .2H 2 O ; moreover, 
 his water determinations, 5.52, 5.72, 5.89, 5.93, and 5.16 per 
 cent, agree satisfactorily with the calculated amount, 5.90, for 
 this salt. 
 
 Remsen and Herty made only a single chloride, and likewise 
 only one bromide. The other chloride, and the two bromides 
 belonging to other types crystallize beautifully and are as 
 easily made as the salts which they prepared, and it is a strange 
 coincidence that the latter happened to correspond in type to 
 the iodide which they had obtained. I have confirmed the 
 composition of their bromide, KPbBr 3 .H 2 O, but their chloride, 
 to which they gave the formula KPbCl 2 is evidently identi- 
 cal with the compound which I have found to be undoubtedly 
 hydrous, 3KPbCl 8 .H 2 O. 
 
 Lowig, as already mentioned, has described the compound 
 K 2 PbBr 4 . I have been unable to find his original article, but 
 from the fact that I have not obtained an anhydrous form of 
 this compound, I believe that he overlooked the water of 
 crystallization or dehydrated the salt before analyzing it. 
 
 A bromide, K 2 Pb 3 Br 8 is mentioned by Berthelot.f He does 
 not give any analysis or description of it, and I am convinced 
 from my own experiments that he obtained a mixture of 
 KPbBr s 4H 2 O and KPb 2 Br 6 . 
 
 Strohecker f states that he produced three different chlorides 
 of potassium and lead by mixing potassium chloride and lead 
 nitrate solutions. It is remarkable, considering the abundance 
 
 * Zeitschr. Physikal. Chem., ix, 57, 1892. 
 t Ann. China. Phys., V, xxix, 289, 1883. 
 t Jahresbericht, 1869, 282. 
 
POTASSIUM-LEAD HALIDES. 253 
 
 and cheapness of the materials and the ease with which large 
 quantities of the double salts can be made, that he did not 
 obtain them in sufficient quantities for exact analyses. Since 
 I have succeeded in making only two double chlorides, I believe 
 that one of Strohecker's salts, which he describes as feathery, 
 was simply lead chloride. 
 
 The results of previous investigators may be summed up 
 by saying that it is probable that no potassium-lead halides have 
 been correctly described, if water of crystallization is taken 
 into consideration, except two of Remsen and Herty's salts, 
 KPbBr 3 .H 2 O and KPbI 3 .3H 2 O. 
 
 Method of Preparation. 
 
 Both the caesium and potassium salts have been investi- 
 gated, in every case, by making hot, aqueous solutions of the 
 component halides and cooling to crystallization. Some pre- 
 vious investigators had used solutions of lead nitrate and an 
 alkaline halide for the purpose, but their example has not been 
 followed, because it was not believed that the presence of an 
 alkaline nitrate would in any way facilitate the operation, and 
 it was feared that it might incur contamination in some cases. 
 The conditions were gradually varied from a point where the 
 alkaline halide crystallized out, to a point where the lead 
 halide was deposited uncombined, and the experiments were 
 so carefully earned out and so frequently repeated that it 
 seems scarcely possible that any double salt was overlooked. 
 
 The salts have been made on a rather large scale. In the 
 case of the caesium compounds, the rarity of the material made 
 it necessary to perform the separate experiments with only 
 about 50 or 75 g. of a caesium halide, but in making the 
 potassium salts 400 or 500 g. of a potassium halide were 
 frequently used. 
 
 Solutions which were neutral or slightly acid were generally 
 used. The effect of the presence of a large amount of free 
 acid, hydrochloric, hydrobromic, or hydroiodic, as the case re- 
 quired, was also carefully studied, but these had no apparent 
 effect upon the results. 
 
254 ON THE CESIUM- AND THE 
 
 Very large crops of the potassium salts were sometimes 
 formed, so that the homogeneity of the mass was doubtful. 
 In such cases the greater part of the crop was removed and 
 satisfactory crystals were obtained by dissolving the remainder 
 in the mother-liquor by the aid of heat and cooling. 
 
 The caesium material used was wholly from the pollucite of 
 Hebron, Maine.* The salts were carefully purified for this 
 investigation. Godeffroy's method f was found to be very 
 satisfactory for the purpose of separating caesium from the 
 sodium and potassium which accompany it in the mineral. 
 
 Kahlbaum's potassium chloride, bromide, and iodide were 
 usually used for making the potassium salts, but for a few 
 experiments the ordinary medicinal potassium bromide was 
 substituted. Since some of the analyses of the double bro- 
 mide show an excess over 100 per cent, it is suspected that 
 the salts contained a little chlorine. Calculation shows that 
 one per cent of chlorine replacing bromine would cause an 
 excess of 0.71 per cent if the chlorine was weighed as silver 
 chloride and calculated as bromine. 
 
 The lead halides which were used were prepared by our- 
 selves from reliable materials. 
 
 General Properties. 
 
 The lead double halides are all decomposed by water, and 
 the presence of a large excess of the alkaline halide is neces- 
 sary for the formation of all the compounds to be described 
 except CsPbjjCls and CsPb 2 Br 6 , which are almost stable with 
 water. The concentration of the alkaline halide solution 
 evidently determines, in the cases of the chlorides and bro- 
 mides, the type of salt produced. Since the simple caesium 
 halides are much more soluble than those of potassium, it is 
 possible to use them in much more concentrated solutions, 
 and the salts of Cs.PbCle and Cs 4 PbBr 6 are readily obtained. 
 In the case of potassium bromide the solution becomes satu- 
 rated with the simple salt by concentration just beyond the 
 
 * Amer. Jour. Sci., Ill, xli, 213. f Berichte d. Chem. Ges., vii, 375. 
 
POTASSIUM-LEAD HALIDES. 255 
 
 point where K 2 PbBr 4 .H 2 O is obtained, and with potassium 
 chloride, which is less soluble than the bromide, the limit is 
 reached at the compound 3KPbCl 3 .H 2 O. The apparent exist- 
 ence of only a single double iodide, both with caesium and 
 potassium, is remarkable, since caesium iodide is very soluble 
 and potassium iodide is much more soluble than the bromide 
 and chloride. 
 
 On account of their decomposition by water, no determina- 
 tions of the solubility of the double halides have been made, 
 but it was noticed that the caesium compounds were much less 
 soluble in the saline solutions than the corresponding potas- 
 sium salts. This relation corresponds with the observation of 
 Godeffroy,* that while the simple salts increase in solubility 
 from potassium to caesium, the double and complicated salts 
 show a decrease in this direction. 
 
 All the chlorides and bromides described in this article are 
 colorless, or in one case nearly so except two caesium salts, 
 CsPbCls and one modification of CsPbBr 8 . The first of these 
 is pale yellow and the other bright orange. These colors are 
 very remarkable since the simple halides from which they are 
 made are all colorless. I have previously observed a similar 
 case, where a colored double halide was formed from two 
 colorless halides, in the compound CsHgBr 8 . f Both double 
 iodides are yellow, the hydrous potassium salt being the paler 
 of the two. 
 
 Analytical Methods. 
 
 Great care was used in selecting homogeneous material for 
 analysis. The crystals were dried as rapidly and thoroughly 
 as possible by pressing them between smooth filter-papers, and 
 where the substance did not lose its lustre by the operation, 
 it was then exposed to the air for several hours. 
 
 Water was determined by collecting and weighing it in a 
 calcium-chloride tube, the substance being ignited in a com- 
 bustion-tube, behind a layer of dry sodium carbonate, in a 
 
 * Berichte d. Chem. Ges., ix, 1365. t Amer. Jour. Sci., Ill, xliv, 227. 
 
256 ON THE CESIUM- AND THE 
 
 current of dry air. The water lost over sulphuric acid or at 
 certain temperatures was determined by the usual methods. 
 
 Lead was determined in two ways. With all the caesium 
 salts the substance was dissolved in hot water (an easy opera- 
 tion with all these salts, but impracticable in the case of some 
 of the potassium compounds), and all except a trace of lead 
 was precipitated by ammonium carbonate in presence of am- 
 monium hydroxide. The precipitate of lead carbonate was 
 removed by filtration, and the remaining trace of lead was pre- 
 cipitated by passing hydrogen sulphide into the alkaline solu- 
 tion. The lead sulphide was collected and ignited by itself in 
 a porcelain crucible. The amount of this was so small that 
 it was evident that no appreciable error would arise from any 
 lead sulphate that the ignited residue might contain, so that 
 the main precipitate of lead carbonate was ignited in the same 
 crucible and the whole was weighed and calculated as lead 
 oxide. A different method was selected for the determination 
 of lead in the potassium compounds, for the reason that some 
 of them could not be readily dissolved in hot water, and it was 
 found to be more convenient and expeditious than the other. 
 About 1 g. of substance was dissolved in about 10 c. c. nitric 
 acid (sp. gr. 1.20), about 2c.c. concentrated sulphuric acid, 
 previously diluted with water, were then added, and the nitric 
 acid was removed by evaporation. After diluting with about 
 25 c. c. of water and cooling, the lead sulphate was collected 
 in a Gooch crucible, washed with very dilute sulphuric acid, 
 ignited, and weighed. 
 
 In order to determine csesium, the alkaline solution from 
 which the lead had been removed was concentrated until the 
 ammonium carbonate, hydroxide, and sulphide had been nearly 
 or quite removed, a small excess of sulphuric acid was added, 
 and, after evaporation and ignition, normal csesium sulphate 
 was obtained by igniting in a current of air containing am- 
 monia, and this was weighed. 
 
 The nitrates from the lead sulphate did not contain an 
 appreciable amount of lead. Normal potassium sulphate was 
 obtained from these solutions by evaporating, igniting, and 
 heating in an ammoniacal atmosphere. 
 
POTASSIUM-LEAD HALIDES. 257 
 
 The halogens were determined as silver halides. Where 
 the substance could be completely dissolved in hot water, an 
 excess of silver nitrate was added to the hot solution and it 
 was afterwards acidified with nitric acid. When it happened 
 that the lead halide remained partly undissolved, the nitric 
 acid was not added until this had been completely decomposed 
 by long digestion on the water-bath with an excess of silver 
 nitrate. The precipitates were collected and weighed in Gooch 
 crucibles. 
 
 THE CAESIUM-LEAD CHLORIDES. 
 
 BY G. F. CAMPBELL. 
 
 CstPb C1 6 . When lead chloride is dissolved, by the aid of 
 heat, in a solution of caesium chloride which is so concentrated 
 as to be nearly saturated when cold, this salt is deposited on 
 cooling in the form of brilliant white rhombohedrons. Crys- 
 tals having a diameter of 2 or 3 mm. were sometimes obtained. 
 Two entirely separate crops were analyzed, both of which were 
 undoubtedly free from other compounds. 
 
 For 
 
 Caesium .... 55.60 
 Lead 
 
 md. 
 
 56.03 
 21.63 
 22.23 
 
 Calculated for 
 
 C8 4 PbCl. 
 
 55.90 
 21.75 
 22.35 
 
 Chlorine .... 21.97 
 
 99.89 
 
 100.00 
 
 CsPl Cl s . On gradually diluting the concentrated solution 
 of csesium chloride, such as was used in making the previous 
 salt, and dissolving lead chloride in it as before, a point is soon 
 reached where short prismatic crystals of small size and of a 
 pale yellow color are deposited on cooling. Three different 
 crops of apparently pure crystals were analyzed. 
 
 Caesium 
 
 Lead 
 
 Chlorine 
 
 
 Pound. 
 
 
 Calculated for 
 CsPbCl s . 
 
 29.79 
 
 31.33 
 
 30.54 
 
 30.13 
 
 44.99 
 
 45.28 
 
 46.29 
 
 46.36 
 
 23.85 
 
 23.75 
 
 23.71 
 
 23.85 
 
 100.17 
 
 99.57 
 
 100.13 
 
 100.00 
 
 17 
 
258 
 
 ON THE CESIUM- AND THE 
 
 ^ 01 B . Experiments with still more dilute solutions, 
 carried out in a similar manner, gave, under wide variations 
 of conditions, this salt in the form of thin white plates which 
 were often several millimeters in diameter. These plates pre- 
 sented marked variations in habit, which were apparently due 
 to changes in the conditions under which they were made. 
 In two crops, of which A and B are the analyses, the plates 
 were uniformly rhomboidal in form. Two other crops, C and 
 D, were made up of lengthened plates, so twinned as to form 
 feathery aggregates. In another crop, E, made from a more 
 dilute solution than the others, the plates were apparently 
 square. 
 
 Caesium 
 
 Lead 
 
 Chlorine 
 
 
 
 
 
 
 Calculated for 
 CsPb,Cl 5 . 
 
 18.36 
 
 19.99 
 
 B. 
 
 18.44 
 
 c. 
 18.27 
 
 D. 
 
 E. 
 
 18.45 
 
 57.14 
 
 57.16 
 
 57.06 
 
 56.98 
 
 57.08 
 
 57.16 
 
 . . . 
 
 24.47 
 
 . , . 
 
 24.52 
 
 24.35 
 
 24.48 
 
 100.07 
 
 99.88 100.00 
 
 The three different habits in which this salt crystallizes are 
 so distinct in appearance that, before the samples were ana- 
 lyzed, it was supposed that they were separate compounds. It 
 .appears probable that the compound is at least dimorphous. 
 
 THE CAESIUM-LEAD BROMIDES. 
 
 BY P. T. WALDEN. 
 
 CsiPbBrs. This salt is produced, in concentrated solutions, 
 similarly to the corresponding chloride. Like the latter salt, it 
 forms white rhombohedrons. The crystals were usually not 
 over 1 or 2 mm. in diameter. Two separate crops were prepared 
 and analyzed. 
 
 Found. 
 
 Caesium .... 43.61 43.42 
 Lead .... 16.83 16.83 
 Bromine 39.24 39.33 
 
 Calculated for 
 Cs 4 P'bBr 6 . 
 
 43.64 
 16.98 
 39.38 
 
 99.68 99.58 
 
 100.00 
 
 CsPbBr s . This compound is dimorphous. One modifica- 
 tion forms small prisms of a bright orange color, the other is 
 
POTASSIUM-LEAD HAL1DES. 259 
 
 pure white and crystallizes in slender needles. The orange 
 salt is obtained when lead bromide is dissolved in somewhat 
 more dilute solutions of caesium bromide than those required 
 for the formation of Cs 4 PbBr 6 , and there is a narrow range of 
 conditions where it crystallizes upon the latter salt. There is, 
 therefore, no evidence of the existence of an intermediate 
 compound, Cs a PbBr 4 , corresponding to one of the potassium- 
 lead bromides. Whenever solid lead bromide is added to a 
 concentrated solution of caesium bromide, it instantly loses its 
 white color and takes on that of the orange salt. The white 
 needles are formed in solutions which are slightly more dilute 
 than those required for the orange modification. The limits 
 of the conditions under which this white salt is formed are 
 very narrow, and a great many trials were necessary before 
 satisfactory crops were obtained. Two distinct samples of 
 each salt were analyzed. The white needles were not abso- 
 lutely free from the orange compound, but there is no doubt 
 that they were sufficiently pure to show their composition 
 accurately. 
 
 Caesium . . . 
 Lead .... 
 Bromine . . . 
 
 100.25 99.86 99.73 99.82 100.00 
 
 On heating the white modification to about 140, it gradu- 
 ally assumes the exact color of the orange salt, without chang- 
 ing its external form, and this color is permanent on cooling. 
 
 CsPb 2 Br B . This salt is produced in solutions which are 
 still more dilute than those from which the preceding com- 
 pounds are obtained. It was first noticed at a volume of 
 about 160 c. c. of a solution containing about 50 g. of caesium 
 bromide. It continued to form, on further dilution and the 
 addition of lead bromide, until the volume reached 1250 c. c., 
 when lead bromide began to be deposited. The conditions 
 under which the salt is formed are, therefore, very wide. The 
 compound crystallizes in thin white plates, which, like 
 
 
 
 Calculated for 
 CsPbBr 3 . 
 
 22.93 
 35.69 
 41.38 
 
 Orange Salt. 
 
 23.19 23.13 
 35.69 35.39 
 41.37 41.34 
 
 White Salt. 
 
 23.02 22.49 
 35.24 35.88 
 41.47 41.45 
 
260 ON THE CAESIUM- AND THE 
 
 the corresponding chloride, present considerable differences 
 in habit. Plates having a diameter about 5 mm. were 
 sometimes obtained. Three separate crops of crystals were 
 analyzed. 
 
 Pound. Calculated for 
 * * CsPb 2 Br 6 . 
 
 Osium . . 14.13 14.35 
 Lead . . . 43.39 43.72 43.45 
 Bromine . . 42.23 42.21 
 
 99.75 100.28 
 
 THE CAESIUM-LEAD IODIDE AND SOME MIXED DOUBLE 
 HALIDES. 
 
 BY A. P. WHEELER. 
 
 CsPlIy Under a great variety of conditions this was the 
 only double iodide that could be produced. The compound is 
 but slightly soluble in hot caesium-iodide solutions, so that the 
 crops obtained were always small. It forms very slender 
 rectangular prisms which are yellow in color. The following 
 analyses were made on separate products : 
 
 Found. 
 
 Calculated for 
 
 CsPbI s . 
 
 Caesium .... 17.90 . . . 18.45 
 
 Lead .... 28.38 27.40 28.71 
 
 Iodine .... 52.83 52.57 52.84 
 
 99.11 100.00 
 
 Three double salts have been made by dissolving lead 
 bromide in solutions of caesium chloride. The analyses show 
 that the two salts do not combine unchanged, but that there 
 is usually an extensive exchange of halogens. Each of the 
 products must be considered, therefore, as a mixture of a 
 double chloride with the corresponding double bromide. 
 
 (7s 4 P6( 01, Br)t. This was produced in rhombohedrons, like 
 the chloride and bromide. Two crops were analyzed. 
 
POTASSIUM-LEAD HALIDES. 261 
 
 Found. 
 
 Caesium 54.65 55.50 
 
 Lead 19.30 18.61 
 
 Chlorine .... 15.89 19.90 
 
 Bromine .... 9.52 4.03 
 
 99.36 98.04 
 
 Katio Br : Cl . . . 1 : 3.8 1 : 11.2 
 
 CsPb( Cl, r) 3 . This occurred in small rectangular prisms, 
 like the chloride and bromide and having a yellow color in- 
 termediate between them. Two crops gave the following 
 analyses : 
 
 Found. 
 
 Caesium 30.24 30.50 
 
 Lead ...... 44.23 43.55 
 
 Chlorine 21.44 18.94 
 
 Bromine 4.00 8.79 
 
 99.91 101.96 
 
 Katio Br : Cl . . . 1 : 12 1 : 4.8 
 
 CsPbz(CI 9 Br)i. This was obtained in white plates resem- 
 bling the two double salts. Two products were analyzed. 
 
 Found. 
 
 Caesium 18.94 
 
 Lead 51.40 51.97 
 
 Chlorine 16.29 19.31 
 
 Bromine 13.27 8.62 
 
 99.90 
 
 Eatio Br : Cl 1 : 2.8 1:5 
 
 THE POTASSIUM-LEAD HALIDES. 
 
 In studying these bodies care has been taken to record the 
 conditions under which they were made. These conditions in 
 many cases are only approximately given, because uncertain 
 quantities of salts had often been removed from the solutions, 
 either for analysis or in order to obtain smaller and better 
 crops of crystals. A large number of analyses have been 
 
262 
 
 ON THE CESIUM- AND THE 
 
 made in some cases. This was due to the fact that the salts 
 often varied so little in appearance that it was necessary to 
 analyze many products in order to identify them and to be 
 certain that they were not different compounds. 
 
 3KPbOl s .ff s O. When lead chloride is dissolved in a hot 
 solution of potassium chloride which is so concentrated as to 
 be nearly saturated when cold, this double salt is deposited on 
 cooling. It forms brilliant prismatic crystals which are largest 
 in the most concentrated potassium-chloride solutions. The 
 largest crystals obtained had a length of more than 10 mm. 
 and a diameter of 1 or 2 mm. It was noticed that, when suffi- 
 ciently concentrated solutions were used, pure potassium 
 chloride crystallized upon this compound, and no evidence 
 was obtained of the existence of a double salt containing a 
 larger proportion of potassium chloride than this. 
 
 The following table gives the approximate conditions under 
 which the five samples which were analyzed were made : 
 
 A 
 B 
 C 
 
 D 
 E 
 
 The results of the analyses are as follows : 
 
 Pound. 
 
 KC1. 
 
 PbCl 2 . 
 
 Volume. 
 
 Volume for 
 1 g. KCL 
 
 g- 
 
 g- 
 
 c. c. 
 
 c. c. 
 
 400 
 
 30 
 
 1100 
 
 2| 
 
 400 
 
 80 
 
 1200 
 
 3 
 
 150 
 
 40 
 
 450 
 
 3 
 
 100 
 
 25 
 
 350 
 
 3 
 
 300 
 
 55 
 
 1300 
 
 H 
 
 K . . 
 
 A. 
 
 . 11.38 
 
 B. 
 
 11.10 
 
 c. 
 10.79 
 
 D. 
 
 E. 
 
 3KPbCl 3 .H 2 0. 
 
 1090 
 
 Pb . 
 
 Cl . . 
 
 . 57.46 
 . 29.91 
 
 57.68 
 29.87 
 
 57.43 
 29.81 
 
 57.94 
 
 57.14 
 
 57.73 
 29.70 
 
 H 2 . 
 
 . 1.45 
 
 1.39 
 
 
 1.51 
 
 1.88 
 
 1.67 
 
 
 100.20 
 
 100.04 
 
 
 
 
 100.00 
 
 All the samples were thoroughly air-dried before they were 
 analyzed. By this treatment the crystals did not lose any of 
 their lustre. A finely pulverized portion of sample A lost 
 only 0.02 per cent in weight after standing over concentrated 
 
POTASSIUM-LEAD HALIDES. 263 
 
 sulphuric acid for eight days. The same sample suffered an 
 additional loss of 0.23 per cent when heated for twelve hours 
 in a steam drying-oven. The water was not rapidly given off 
 until a temperature of about 200 was reached. The salt 
 decrepitates when heated rapidly to about 200, corresponding 
 in this respect to the salt which Remsen and Herty described 
 as anhydrous and to which they gave the formula KPbCl s . 
 There can be no doubt, therefore, that Remsen and Herty's 
 formula is incorrect. 
 
 KPb z C1 5 . This salt is formed in more dilute solutions than 
 those which produce the previously described compound. It 
 occurs, like that compound, in white prismatic crystals, but it 
 differs considerably from it in lustre and form, so that the two 
 salts can be distinguished by microscopic examination. The 
 salt under consideration is anhydrous, and this fact makes it 
 easy to distinguish this compound, when pure, from the other. 
 
 Four analyzed crops were made under the following condi- 
 tions : 
 
 Volume for 
 1 g. KC1. 
 
 g. g. c. c. 
 
 A 
 B 
 
 C 150 20 1100 7 
 
 D 
 
 The analyses were as follows : 
 
 Pound - Calculated for 
 
 KC1. 
 
 PbCl,. 
 
 Volume. 
 
 200 
 
 g- 
 
 50 
 
 c.c. 
 
 1500 
 
 150 
 
 30 
 
 1100 
 
 150 
 
 20 
 
 1100 
 
 250 
 
 55 
 
 1200 
 
 Potassium . . 
 Lead .... 
 
 A. 
 
 . 6.14 
 . 64.74 
 
 B. 
 
 5.97 
 66.43 
 
 c. 
 6.18 
 65.85 
 
 D. 
 
 6.07 
 65.72 
 
 KPb 2 Cl 8 
 
 6.20 
 65.65 
 
 Chlorine . . 
 Water . . . 
 
 . 28.11 
 . 0.11 
 
 
 28.13 
 
 28.08 
 
 28.15 
 0.00 
 
 99.10 100.16 99.87 100.00 
 
 There was no indication of the formation of any other 
 double chloride, as the dilution was increased beyond that 
 given for the above products, and when a solution containing 
 1 g. of KC1 in 11 c. c. was used pure lead chloride was deposited. 
 H^O This salt is obtained by dissolving lead 
 
264 
 
 ON THE CAESIUM- AND THE 
 
 bromide in the most concentrated solutions of potassium 
 bromide. It forms brilliant prismatic crystals which are 
 permanent in the air. The largest of these which were ob- 
 tained were about 1 mm. in diameter and 5 mm. in length. A 
 number of crops were made under the following conditions : 
 
 A 
 B 
 C 
 
 D 
 E 
 F 
 
 KBr. 
 400 
 
 400 
 400 
 400 
 500 
 500 
 
 PbBr 2 
 g- 
 
 70 
 90 
 120 
 130 
 130 
 130 
 
 Volume, 
 c. c. 
 
 700 
 700 
 800 
 650 
 
 850 
 
 775 
 
 Volume for 
 1 g. KBr. 
 
 If 
 If 
 
 2 
 
 These products gave the following analyses : 
 
 A 
 
 K. 
 
 12.51 
 
 Pb. 
 
 34.25 
 
 Br. 
 
 51.47 
 
 H 2 0. 
 
 2.50 
 
 = 100.73 
 
 B 
 
 1221 
 
 34.59 
 
 51.21 
 
 2.51 
 
 - 100.52 
 
 c 
 
 11.89 
 
 34.47 
 
 51.14 
 
 2.44 
 
 = 99.94 
 
 D 
 
 12.37 
 
 34.50 
 
 51.35 
 
 
 
 E 
 
 
 34.26 
 
 51.40 
 
 2.61 
 
 
 
 1270 
 
 3389 
 
 51.46 
 
 2.57 
 
 - 100.62 
 
 Calculated for 
 K 2 PbBr 4 .H 2 
 
 
 33.21 
 
 51.35 
 
 2.89 
 
 = 100.00 
 
 This salt is apparently stable in the air, but it loses water 
 very slowly over sulphuric acid. A finely powdered sample of 
 A lost 0.23 per cent after remaining twelve hours in the desic- 
 cator, and the same portion suffered an additional loss 0.33 
 after eight days. A sample which was not pulverized lost only 
 0.09 per cent in twelve hours and, in addition, 0.17 per cent in 
 eight days. About one-half of the water went off when the 
 substance was heated for twelve hours in a steam drying-oven. 
 At 200 the water is rapidly and completely expelled. 
 
 QKPbBrs.Hz 0. The conditions under which this salt can 
 be made are rather narrow, and these conditions encroach upon 
 those of the preceding compound, so that small differences in 
 the amounts of lead chloride used or in the temperature of 
 the solution are sufficient to cause the formation of the other 
 
POTASSIUM-LEAD HALIDES. 
 
 265 
 
 salt. It forms brilliant, colorless, lozenge-shaped crystals 
 which can be easily distinguished from the other compound. 
 The crystals which were obtained sometimes had a diameter of 
 2 or 3 mm. 
 
 The crops analyzed were made under the following con- 
 ditions : 
 
 A 
 
 B 
 C 
 D 
 
 E 
 
 The analyses were as follows : 
 
 KBr. 
 
 PbBr,. 
 
 Volume. 
 
 g. 
 
 500 
 
 g- 
 
 130 
 
 c. c. 
 
 950 
 
 500 
 
 130 
 
 1050 
 
 500 
 
 140 
 
 900 
 
 500 
 
 120 
 
 1050 
 
 500 
 
 120 
 
 1125 
 
 Volume for 
 1 g. KBr. 
 
 2* 
 
 A 
 
 K. 
 
 , 8.44 
 
 Pb. 
 
 41.91 
 
 Br. 
 
 H 2 0. 
 
 129 
 
 
 B . . . . , 
 
 , 802 
 
 4271 
 
 4895 
 
 1 62 
 
 101 30 
 
 C . . . . 
 
 , 8.60 
 
 41.61 
 
 49.16 
 
 1.60 
 
 10097 
 
 D 
 
 , 8.08 
 
 42.69 
 
 48.91 
 
 1 14 
 
 10082 
 
 E 
 
 
 4261 
 
 
 117 
 
 
 Calculated for 
 3KPbBr 8 .H 2 O 
 
 i 7.95 
 
 42.06 
 
 48.77 
 
 1.22 = 
 
 100.00 
 
 The salt is stable in the air. A sample, after standing 
 seven days over sulphuric acid, lost only 0.04 per cent. The 
 water is given off very slowly at 100. 
 
 KPbBr z .H<i 0. This salt was described by Remsen and 
 Herty. At summer temperature, about 25, I was unable to 
 obtain it, but by placing the mother-liquors from the preced- 
 ing salt in an ice-chest, beautifully crystallized crops of it 
 were obtained. 
 
 Its formation was also noticed at laboratory temperatures 
 when the weather was somewhat cooler than in midsummer. 
 It forms prismatic crystals. Some of those obtained were 
 about 10 mm. long and 2 mm. in diameter. Two crops were 
 analyzed. 
 
266 ON THE CAESIUM- AND THE 
 
 , Calculated for 
 
 KPbBr 3 .H 2 0. 
 
 Potassium 8.24 7.90 7.76 
 
 Lead 41.23 41.20 41.06 
 
 Bromine 47.81 . . . 47.61 
 
 Water 3.28 3.64 3.57 
 
 100.56 100.00 
 
 The salt is usually permanent in the air, but in dry weather 
 the crystals gradually become opaque, and over sulphuric 
 acid about two-thirds of the water is rapidly given off. 
 
 KPb z Br 5 . This salt crystallizes in square plates, sometimes 
 3 or 4 mm. in diameter. It can be readily distinguished from 
 the other double bromides, not only by its form, but from the 
 fact that it quickly assumes a pale green color when exposed 
 to daylight. On long exposure, or in direct sunlight, this 
 color changes to a pale dirty-brown. I have observed that 
 lead bromide itself becomes nearly black on long exposure to 
 daylight. This fact does not appear to be generally known. 
 
 The samples analyzed were made under the following 
 conditions : 
 
 KBr. PbBr,. Volume. 
 
 A 400 130 1050 
 
 B 400 150 1250 
 
 C 200 75 1000 
 
 The results of the analyses are as follows : 
 
 Found. 
 
 
 A. 
 
 B. 
 
 c. 
 
 Potassium . 
 
 . 4.75 
 
 4.75 
 
 4.71 
 
 Lead . . . 
 
 . 49.22 
 
 49.11 
 
 48.48 
 
 Bromine . . 
 
 . 47.03 
 
 46.98 
 
 46.89 
 
 
 101.00 
 
 100.84 
 
 100.08 
 
 Calculated for 
 KPb 2 Br 8 . 
 
 4.58 
 
 48.53 
 46.89 
 
 100.00 
 
 . It has already been mentioned that this is 
 the only double iodide that either Remsen and Herty or I 
 have been able to make. It forms slender, pale yellow needles, 
 and is produced under a great variety of conditions. Two 
 samples were analyzed. A was made with about 450 g. KI, 
 
POTASSIUM-LEAD HALIDES. 267 
 
 75 g. PbI 2 , and 600 c. c. volume. For B about 400 g. KI, 
 45 g. PbI 2 , and 280 c. c. volume were used. 
 
 Found. Calculated for 
 
 A. B. KPbI s .2H,0. 
 
 Potassium 6.03 6.07 5.90 
 
 Lead 30.73 30.13 31.21 
 
 Iodine 57.57 56.99 57.46 
 
 Water 5.26 6.04 5.43 
 
 99.59 99.23 100.00 
 
 The salt is apparently stable in the air, but it loses water in 
 the desiccator. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 October, 1892. 
 
ON THE DOUBLE HALIDES OF TELLURIUM WITH 
 POTASSIUM, RUBIDIUM, AND CESIUM.* 
 
 BY H. L. WHEELER. 
 
 THE existence of double halides of tellurium with potas- 
 sium, sodium, and ammonium was first indicated by Berzelius.f 
 He described the methods by which he obtained them, but 
 gave no analyses of the compounds. Later, RammelsbergJ 
 investigated the double chlorides of tellurium with potassium 
 and ammonium, with the object of determining their composi- 
 tion. He arrived at the formula? 8KC1.3TeCl 4 and 8NH 4 C1. 
 3TeCl 4 . It will be shown beyond that the formula of the 
 potassium compound at least must have been obtained from 
 analyses of impure products. Von Hauer analyzed the 
 double bromide of tellurium and potassium, and concluded 
 that the salt had the composition represented by the formula 
 2KBr.TeBr 4 .3H 2 O. I have reinvestigated this salt and 
 found it to contain two molecules of water and not three. 
 Probably Von Hauer analyzed the salt without previously 
 having dried it sufficiently or without haying taken precau- 
 tions to remove included water which the crystals always con- 
 tain. He dehydrated this salt and used it in his work on 
 the atomic weight of tellurium. 
 
 More recently Wills || determined the atomic weight of 
 tellurium by means of the same salt. He does not give any 
 analyses of the hydrous compound, but states that the salt 
 contains water and gives directions for dehydrating it. Ram- 
 melsberg in his " Handbuch der krystallographisch-physi- 
 kalischen Chemie" (p. 289) quotes the formula of the 
 
 * Amer. Jour. Sci., xlv, April, 1893. t Pogg. Ann., xxxii, 577. 
 t Berlin Monats. Ber., 1875, 379. J. prakt. Chem., Ixxiii, 98. 
 
 || Jour. Chem. Soc., xxxv, 711. 
 
DOUBLE HALIDES OF TELLURIUM. 269 
 
 dehydrated compound from Wills's work and assigns to this 
 Baker's * measurements, which do not belong to it, but to the 
 hydrated compound with the three supposed molecules of 
 water of crystallization. The present investigation has shown 
 that the anhydrous salt is isometric, the hydrous one being 
 orthorhombic. 
 
 Ramsay f says that " By mixing aqueous solutions of 
 the constituent halides, tellurium halides combine thus : 
 TeCl 4 2KCl, TeBr 4 .2KBr, TeI 4 .2KI. These compounds form 
 reddish crystals. Few attempts have been made to prepare 
 double halides." Although a thorough search of the litera- 
 ture on this subject has been made, in connection with the 
 present work, no analyses of the double chloride or iodide 
 could be found. Berzelius's work as regards their prepara- 
 tion and Rammelsberg's attempt to determine the formula of 
 the chloride comprise all the work that has been done on these 
 two salts. It must be concluded that the formulae given by 
 Ramsay were deduced by analogy with the double bromide, 
 especially since his statements in regard to color, method of 
 preparation and composition only apply, in all respects, to the 
 double bromide. 
 
 It will be seen from the above summary that very little 
 satisfactory work has been done on this class of compounds, 
 and therefore the present investigation has been undertaken 
 with the view of making a thorough study of the double 
 halides of tellurium with potassium, rubidium, and csesium. 
 As a result the following compounds have been prepared : 
 
 2KCl.TeCl 4 2RbCl.TeCl 4 2CsCl.TeCl 4 
 
 2KBr.TeBr 4 2RbBr.TeBr 4 2CsBr.TeBr 4 
 
 2KBr.TeBr 4 .2H 2 2RbI.TeI 4 2CsLTeI 4 
 
 2KI.TeI 4 .2H 2 
 
 It is to be noticed that all of these compounds conform to 
 the usual type of double halides of tetravalent metals in con- 
 
 * Jour. Chem. Soc., xxxv, 711. 
 
 t System of Inorganic Chemistry, edition of 1891, p. 168. 
 
270 DOUBLE HALIDES OF TELLURIUM WITH 
 
 taining the alkali metal and tellurium in the ratio of two 
 atoms of the former to one of the latter, and no indications of 
 the formation of salts of a different type were observed. The 
 anhydrous double halides of tellurium crystallize in the isomet- 
 ric system with an octahedral habit, and it is an interesting 
 fact that this form seems to be characteristic for anhydrous 
 double halides of this type. The caesium and rubidium salts 
 are new compounds, as well as the crystallized, anhydrous, 
 double potassium bromide. New formulas have been assigned 
 to the hydrous potassium double bromide and to the double 
 iodide of potassium. A considerable difference is shown in 
 the affinity of the double halides of tellurium and potassium 
 for water of crystallization. The double chloride is anhydrous 
 and no hydrous form of it was observed, the double bromide 
 was prepared in both the hydrous and the anhydrous forms, 
 while the iodide was obtained only with water of crystalliza- 
 tion. This water was more firmly held than in the case of 
 the hydrous bromide, as was shown by the fact that it formed 
 from hot solutions and did not as readily effloresce. 
 
 The methods used in the preparation of pure material for 
 this work, and which deserve to be mentioned on account of 
 giving satisfactory results, are given below. The tellurium 
 was obtained by purifying the commercial product by precipi- 
 tation with sulphurous acid, according to the method of 
 Divers and Shimose*.* The halides of tellurium were pre- 
 pared from this material in the usual way. 
 
 Caesium chloride was obtained in a pure state by the 
 method of Godeffroy.f The bromides and iodides were ob- 
 tained in the usual manner from the carbonate, the latter 
 having been prepared from the pure chloride by converting 
 into nitrate, then into oxalate and igniting the latter, as sug- 
 gested by J. L. Smith,J for the conversion of potassium 
 chloride into carbonate. The rubidium was purified by 
 Allen's acid tartrate method. In the case of the potassium 
 salts, Kahlbaum's pure material was used. 
 
 * Jour. Chem. Soc., xlvii, 439. f Ber. d. chem. Ges., rii, 375. 
 
 t Amer. Jour. Sci., II, xvi, 373. Ibid., xxxiv, 367. 
 
POTASSIUM, RUBIDIUM, AND CAESIUM. 271 
 
 The methods by which the double halides were obtained 
 will be given with the description of the salts. 
 
 Method of Analysis. 
 
 The anhydrous salts were removed from the mother-liquor, 
 and, after pressing on filter paper, were dried in the air. The 
 hydrous compounds were rapidly crushed on smooth filter 
 paper, and, as soon as it was certain that no included water 
 was retained by the fragments, they were placed in weigh- 
 ing tubes. Portions of about half a gram were taken for 
 analysis. In order to determine the halogens, silver sulphate 
 was added to the solution of the weighed sample in water 
 containing a little sulphuric acid. The silver halide was 
 washed, ignited, and weighed in the usual manner. After the 
 removal of the excess of silver by means of hydrochloric acid, 
 tellurium was removed with hydrogen sulphide. This separa- 
 tion of tellurium, best in warm solution, has been found to be 
 complete in a few minutes and in a condition that admits of 
 filtration without inconvenience. The sulphide of tellurium, 
 filtered on asbestos in a Gooch crucible, was washed with 
 water containing a little hydrogen sulphide, then treated with 
 a solution of bromine in dilute hydrochloric acid, which 
 readily dissolves the moist sulphide. An excess of nitric acid 
 was then added to this solution and the whole evaporated on 
 the water bath ; the resulting tellurous acid, being transferred 
 to platinum, was ignited and weighed as TeO 2 . The alkali 
 metals were determined by evaporating the filtrate from the 
 tellurium sulphide to dryness, with an excess of sulphuric acid. 
 The residues were then converted into normal sulphate by 
 ignition in a stream of ammonia, as suggested by Kriiss for 
 potassium sulphate. In the case of the hydrous salts, water 
 was determined by heating them in an air bath to constant 
 weight ; the residues were analyzed and found to correspond 
 in composition to the anhydrous salts. The atomic weights 
 used in the calculation of the results were the following: 
 
 Te, 125 ; K, 39.1 ; Eb, 85.5 ; Cs, 133 ; 01, 35.5 ; Br, 80 ; I, 127. 
 
272 DOUBLE HALIDES OF TELLURIUM WITH 
 
 Solubility. 
 
 The salts are all decomposed by water. The double 
 bromides, however, show an interesting difference in their 
 deportment with this reagent. Potassium tellurium bromide 
 dissolves in a small amount of water, but, if an excess of 
 water is added, tellurous acid separates, as has been observed 
 by Wills.* Rubidium tellurium bromide also dissolves in a 
 little hot water completely, the difference being shown on 
 cooling, when a considerable portion of the tellurium sepa- 
 rates as tellurous acid. In the case of the caesium salt 
 both hot and cold water, in large and small amounts, fail to 
 dissolve the salt, the result being immediate decomposition. 
 Only a small part of the tellurium in this case goes into solu- 
 tion. Most of these double salts can be conveniently recrys- 
 tallized from dilute solutions of the corresponding acid. The 
 exceptions are potassium-tellurium chloride, which is decom- 
 posed by this treatment, and caesium-tellurium iodide, which 
 is practically insoluble in hydriodic acid. The fact, first 
 noticed by Godeffroy,f that double halides, containing the 
 metals potassium, rubidium, and caesium, generally decrease in 
 solubility from potassium to caesium, which has frequently 
 been noticed in this laboratory, is again well illustrated by 
 these compounds. For the determination of the solubility of 
 these salts in acids, they were finely powdered, and saturated 
 solutions were then prepared by digesting a mixture of the 
 acid and an excess of the salt for about a week, at ordinary 
 temperature. This was done in a closed flask. Weighed por- 
 tions of these solutions were evaporated to dryness and the 
 residues dried at 100 and weighed. These solubilities were 
 all taken at 22 , and the results are the average of two or more 
 closely agreeing determinations. 
 
 100 parts HC1 100 parts HC1 
 
 Sp. gr. 1.2 dissolve Sp. gr. 1.05 dissolve 
 
 2RbCl.TeCl 4 . . . 0.34 parts 13.09 parts 
 
 2CsCl.TeCl 4 . . . 0.05 " 0.78 " 
 
 * Loc. cit. 
 
 t Ber. d. Chem. Ges., viii, 9. 
 
POTASSIUM, RUBIDIUM, AND CAESIUM. 273 
 
 100 parts HBr 100 parts HBr 
 
 8p. gr. 1.49 dissolve 8p. gr. 1.08 dissolve 
 
 2KBr.TeBr 4 . . . 6.57 parts 62.90 parts 
 
 2RbBr.TeBr 4 . . . 0.25 3.88 " 
 
 2CsBr.TeBr 4 . . . 0.02 0.13 " 
 
 The double tellurium chlorides, described in this article, 
 are more soluble than the bromides, and the bromides more 
 soluble than the iodides. The solubility of these compounds 
 in strong alcohol shows the same gradation as their solubility 
 in acids, the csesium salts being practically insoluble in this 
 menstruum, while the rubidium salts dissolve to a trifling but 
 clearly perceptible extent, and the potassium salts dissolve 
 considerably or are decomposed with separation of the potas- 
 sium halide, or both solution and decomposition take place, 
 according to the salt experimented with. 
 
 The Chlorides. 
 
 The crystals of the three chlorides have a pale yellow color, 
 resembling that of the well-known ammonium phosphomo- 
 lybdate precipitate, the shade becoming somewhat lighter from 
 the caesium to the potassium salts. 
 
 Ccesium Tellurichloride, 20sQLTeCl^ In the preparation 
 of this compound, and also in the preparation of the rubidium 
 and potassium double chlorides, the tellurium tetrachloride is 
 most conveniently made by converting tellurium into tellurous 
 oxide by means of aqua regia, evaporating to dryness to expel 
 nitric acid and then dissolving the residue in hot hydrochloric 
 acid. An aqueous solution of csesium chloride, added to this, 
 produces a precipitate, even in quite dilute solutions. There 
 must be an excess of hydrochloric acid present to prevent the 
 separation of tellurous acid. On boiling and adding more 
 water, if necessary, this precipitate dissolves. The solution, 
 left to cool, deposits small brilliantly lustrous octahedra. It 
 is a general fact, with these double halides, that an excess of 
 one or the other of the constituents does not affect their com- 
 position. This is shown in this particular case by the fact 
 that it can be recrystallized from strong solutions of tellurium 
 
 or of csesium chlorides. 
 
 18 
 
43.44 
 
 43.90 
 
 44.63 
 
 44.04 
 
 20.65 
 
 . . . 
 
 21.41 
 
 20.69 
 
 35.93 
 
 35.14 
 
 . . . 
 
 35.27 
 
 274 DOUBLE HALIDES OF TELLURIUM WITH 
 
 Cs . . . 
 
 Te . . . 
 
 Cl . . . 
 
 This compound is perfectly stable in the air. It does not 
 melt below the boiling-point of sulphuric acid. It can be 
 precipitated from its solution in dilute hydrochloric acid by 
 the addition of concentrated hydrochloric acid. A portion of 
 the salt, finely pulverized, was treated with water at ordinary 
 temperature. This produced a voluminous white precipitate, 
 which was washed with cold water and dried in the air. 
 
 Te . 
 
 Analysis gave 
 
 .... 71.43 
 
 Calculated for 
 H 2 TeO 8 . 
 
 71.43 
 
 H 2 
 O . 
 
 .... 7.52 
 .... 17.76 
 
 10.29 
 18.28 
 
 Cl . 
 
 .... 2.49 
 
 
 Cs 
 
 0.80 
 
 
 The oxygen which was not given off in the form of water 
 on heating the substance was calculated by difference. From 
 the above analysis the conclusion may be drawn, that the pre- 
 cipitate produced by the action of water on this salt is essen- 
 tially tellurous acid, a small amount of oxychloride of tellurium 
 being present. Hot water dissolves some of this tellurous 
 acid, and, on cooling slowly, the anhydride separates in the 
 characteristic form of colorless octahedra. 
 
 Rubidium Tellurichloride, 2RbCl.TeCl^ The preparation 
 of this salt was in every way analogous to that of the caesium 
 tellurium chloride. However, since this salt is far more solu- 
 ble than the corresponding caesium compound, no precipitate 
 was obtained in dilute solutions. The mixture of the hydro- 
 chloric acid solution of the constituents was concentrated by 
 evaporation, and, when cooled, crystals separated. These 
 were in the form of octahedra, somewhat larger than the 
 caesium salt. 
 
 Kb . . . 
 
 Analysis gave 
 
 . . 33.50 33.83 
 
 Calculated for 
 2RbCl.TeCl 4 . 
 
 3359 
 
 Te . . . 
 
 . . 24 34 
 
 2456 
 
 Cl , 
 
 
 41.85 
 
POTASSIUM, RUBIDIUM, AND CESIUM. 275 
 
 This salt remains permanent in the air. From the dilute 
 hydrochloric acid solution, concentrated hydrochloric precipi- 
 tates it unaltered. Water decomposes it, evidently in the 
 same way as the caesium salt. 
 
 Potassium Tellurichloride, 2KOLTeCl^ To prepare this 
 salt in a pure state an excess of tellurium chloride is necessary. 
 
 The analyzed material was obtained by spontaneous evapora- 
 tion of the constituents in a solution of dilute hydrochloric 
 acid, twice as much tellurium chloride being present as re- 
 quired by the formula. Under these conditions it was found 
 to separate in the form of light yellow octahedra, which, 
 under the microscope, were shown to be free from potassium 
 chloride. 
 
 Analysis gave Calculated for 
 
 Ratio. 2KCl.TeCl 4 . 
 
 K 17.37 0.44 18.79 
 
 Te 30.29 0.24 30.03 
 
 Cl 49.47 1.39 51.18 
 
 97.13 
 
 The salt, therefore, has the formula 2KCl.TeCl 4 . The 
 crystals deliquesce somewhat in moist air, and the analyzed 
 material retained a small amount of water, as is shown by the 
 deficiency in the above analysis. It is not probable that the 
 salt contains water of crystallization, for the crystalline form 
 and optical properties show that it is isomorphous with the 
 anhydrous salts. This salt is the most unstable as well as the 
 most soluble of the anhydrous double halides described in this 
 article. It is readily dissolved by dilute hydrochloric acid. 
 Strong hydrochloric acid separates potassium chloride. It 
 therefore cannot be precipitated from its solutions by the addi- 
 tion of strong hydrochloric acid, as in the case of the other 
 chlorides. Alcohol also separates potassium chloride. Water 
 apparently effects the same decomposition as in the case of the 
 caesium and rubidium chlorides. The tendency of potassium 
 chloride to separate along with the salt explains why Rammels- 
 berg's analysis came high in regard to the potassium chloride. 
 His results corresponded to a mixture of two molecules of KC1 
 and three molecules of 2KCl.TeCl 4 . Experiments with the 
 
276 DOUBLE HALIDES OF TELLURIUM WITH 
 
 calculated quantity of the constituents invariably resulted in 
 the separation of potassium chloride or potassium chloride 
 mixed with the yellow 2KCl.TeCl4. Experiments with the 
 method given by Ramsay * for the preparation of this salt, by 
 mixing aqueous solutions of the constituents, resulted in the 
 decomposition of the tellurium chloride, and the resulting 
 white precipitate failed to dissolve until considerable hydro- 
 chloric acid was added. Attempts to prepare the compound 
 by concentrating the mixture of the constituents by the aid 
 of heat invariably resulted in failure. In certain cases, on 
 cooling such solutions, a mass of colorless slender prisms was 
 obtained, which have not yet been investigated. 
 
 The Double Bromides. 
 
 The crystals of the anhydrous bromides have a brilliant red 
 color resembling that of the mineral crocoite. The powders 
 of the salts have a color that is similar to that of a mixture of 
 equal parts potassium bichromate and red lead. The powder 
 of the hydrous bromide has the color of mercuric oxide, but, by 
 loss of water this soon changes to that of the anhydrous salt. 
 
 Caesium Telluribromide, CsBr. TeBr. This double halide 
 can easily be prepared by mixing finely divided tellurium with 
 caBsium bromide in dilute hydrobromic acid, then adding 
 bromine in excess. The presence of free acid is necessary to 
 prevent the separation of tellurous acid. When the tellurium 
 has disappeared, the solution is concentrated by the aid of 
 heat, and, on cooling, bright red crystals of the pure salt are 
 deposited. These are generally somewhat larger than the 
 crystals of the double chloride. 
 
 Analysis gave Calculated. 
 
 Cs . . . 30.90 30.87 30.91 30.54 
 
 Te . . . 14.29 13.60 14.03 14.35 
 
 Br . . . 55.01 . . . 55.32 55.11 
 
 This salt remains unaltered in the air. It can be separated 
 from its solution in dilute hydrobromic acid by the addition 
 
 * Loc. cit. 
 
POTASSIUM, RUBIDIUM, AND CAESIUM. 277 
 
 of concentrated acid. It does not melt below the boiling- 
 point of sulphuric acid. Attempts to prepare a hydrous 
 salt according to the methods used for the preparation of 
 TeBr 4 .2KBr.2HaO were without success. 
 
 Rubidium Telluribromide, 2RbBr.TeBr^. The directions 
 given for the preparation of the corresponding csesium com- 
 pound apply also in the preparation of this salt. If the solu- 
 tions are strong, the compound separates as a bright red 
 precipitate, but if dilute, on concentrating by means of heat 
 or spontaneous evaporation, it crystallizes in brilliant red 
 octahedra. 
 
 Analysis gave Calculated. 
 
 Kb 22.02 22.04 
 
 Te 16.11 
 
 Br 62.07 61.85 
 
 This salt is stable in the air. Like the corresponding caesium 
 salt, this separates from its solutions by the addition of con- 
 centrated hydrobromic acid. When it is dissolved in a little 
 water and the solution is cooled slowly, colorless octahedrons 
 of TeO 2 separate. The latter product was found to be impure, 
 containing a small amount of bromine. On heating, the salt 
 decrepitates slightly and melts at a high temperature. Efforts 
 to prepare a hydrous salt according to the methods used for 
 the preparation of 2KBr.TeBr 4 .2H 2 O were without success. 
 
 Potassium Telluribromides, ^KBr.TeBr* and 2KBr.TeBr^. 
 2H Z 0.* For the preparation of these salts, a mixture of the 
 constituents was made as described in the case of the csesium 
 double bromide. The solution invariably deposited crystals 
 of the anhydrous salt when it had been concentrated by heat, 
 but, by spontaneous evaporation of the filtrate, the hydrous 
 salt was obtained. On recrystallizing either of these salts 
 from water or from dilute hydrobromic acid, the anhydrous 
 salt is obtained when the solution has been saturated by boil- 
 ing and then allowed to cool, but if the solution is left to 
 deposit crystals at ordinary temperature the hydrous modifica- 
 
 * Described by Von Hauer as containing three molecules of water of 
 crystallization. 
 
278 DOUBLE HALIDES OF TELLURIUM WITH 
 
 tion is obtained. The crystals of these different compounds 
 closely resemble each other in color and appearance. The 
 anhydrous variety crystallizes in octahedra modified by the 
 cube. The orthorhombic crystals of the hydrous salt look like 
 distorted crystals of the other. This being the case, and since 
 the crystals of the hydrous compound can be obtained much 
 larger than those of the anhydrous salt, both Von Hauer and 
 Wills selected these for their work, while the more easily 
 obtained anhydrous salt was overlooked. The hydrous salt is 
 readily distinguished from the anhydrous one by its deport- 
 ment on exposure to a dry atmosphere. The latter is stable, 
 but under these conditions the hydrous compound rapidly 
 effloresces, losing its lustre, the faces of the crystals becoming 
 superficially covered with a light reddish yellow and opaque 
 layer of the anhydrous salt. Crystals which have been ex- 
 posed to dry air for several days and were completely covered 
 with this layer were found, on crushing, to remain unaltered 
 in the interior and to have still retained included water in 
 addition to their water of crystallization. This was shown by 
 the fact that the crushed crystals gave a stain of the mother- 
 liquor to filter paper. This property of the hydrous crystals 
 explains why Von Hauer assigned three molecules of water to 
 this salt instead of two. The material for analysis of the 
 hydrous salt was selected from crystals varying in size from 7 
 to 13mm. in diameter. These were very rapidly crushed on 
 smooth filter paper, to remove included water, and immediately 
 corked up in the weighing tube and analyzed. A close exam- 
 ination of the fragments, before and after weighing, gave no 
 evidence of loss of water from the substance by efflorescence. 
 The analyses were from three different crops. 
 
 Analysis gave Calculated for Calculated for 
 
 2KBrTeBr v 2H 2 O. 2KBr.TeBr 4 .3H 2 O. 
 
 K . . . 10.90 11.07 10.73 10.87 10.61 
 
 Te . . . 17.59 17.29 17.46 17.38 16.96 
 
 Br . . . 66.35 66.36 66.34 66.74 65.11 
 
 H 2 . . 5.33 5.53 5.73 5.01 7.32 
 
 These results make it evident that the salt contains two mole- 
 cules of water, and not three as has generally been supposed. 
 
POTASSIUM, RUBIDIUM, AND CESIUM. 279 
 
 The water in this salt was determined by heating it in an air 
 bath to constant weight. The temperature was maintained 
 between 150-160, and finally, to be sure that all the water 
 had been driven off, the residues were analyzed in two cases. 
 
 Analysis gave 
 
 K ..... 11.71 11.52 11.44 
 
 Te ..... 18.29 18.58 18.30 
 
 Br ..... 70.25 70.09 70.26 
 
 Analyses of products obtained by cooling hot saturated 
 solutions gave the following results : 
 
 Found. Calculated for 
 2KBrTeBr 4 . 
 
 K . . . . 11.67 11.70 . . . 11.44 
 
 Te . . . . 18.06 18.30 
 
 Br . . . . 70.24 70.20 69.40 70.26 
 
 The Double Iodides. 
 
 These salts are all black. The powder of the caesium salt 
 is pure black, that of the rubidium and potassium salts is 
 grayish black. 
 
 Ccesium Telluriiodide, 2 Osl. Tel. In order to prepare this 
 salt, and also in the case of the rubidium and potassium com- 
 pounds, tellurium tetraiodide was made by treating tellurous 
 oxide with hydriodic acid. The iodide of tellurium is spar- 
 ingly soluble in hydriodic acid, but, on mixing this solution 
 with a solution of csesium iodide, an amorphous black precipi- 
 tate was obtained, even in very dilute solutions. 
 
 Anai-mn'a . Calculated for 
 
 Analysis gave 2CsLTeI i . 
 
 Cs 23.37 23.07 
 
 Te 10.51 10.84 
 
 I 65.17 66.09 
 
 This compound resisted all attempts to prepare it in a crys- 
 talline form. It is insoluble in csesium iodide and in hydriodic 
 acid ; hence warming in the mother-liquor failed to dissolve the 
 salt. It is decomposed slowly by cold water, rapidly by hot, 
 and apparently tellurous acid or anhydride separates. This 
 generally is impure, being mixed with a dark-colored residue 
 
280 DOUBLE HALIDES OF TELLURIUM WITH 
 
 containing iodine. On exposure the salt slowly loses iodine. 
 In the open capillary it does not melt below the boiling-point 
 of sulphuric acid. 
 
 Rubidium Telluriiodide, 2KbLTeI^ This compound was 
 prepared by mixing the constituents in the same manner as in 
 the preparation of the corresponding caesium salt. If the solu- 
 tions are only moderately concentrated, a black amorphous pre- 
 cipitate is produced. Unlike the corresponding caesium salt, it 
 dissolves, to a slight extent, on warming in the mother-liquor, 
 and on cooling, black microscopic octahedra are produced. 
 
 Analysis gave 
 
 Eb ..... 16.83 16.17 
 
 Te ........ 11.81 
 
 I ..... 72.07 72.02 
 
 This iodide is stable on exposure. Water effects the same 
 decomposition as in the case of the caesium salt. A small 
 portion of this salt dissolves in strong alcohol, giving the color 
 of a weak iodine solution. 
 
 Potassium Telluriiodide, 2KI. TeI^H z 0. This compound 
 can be prepared most conveniently by boiling tellurium iodide 
 in a strong solution of potassium iodide in dilute hydriodic 
 acid. The solution, filtered while hot from any undissolved 
 tellurium iodide, deposits long black prisms on cooling. These 
 crystals attain considerable size, about 30 mm. in length, when 
 a large excess of potassium iodide is used. The mother-liquor, 
 on evaporation in a desiccator, deposits more of the salt, but 
 the crystals have a different habit. 
 
 K . . . 
 Te . . . 
 I ... 
 
 H 2 . . 
 
 For the determination of water in this compound the crystals 
 were rapidly pressed on paper and immediately analyzed. It 
 was found that the salt could be dehydrated at a temperature 
 
 Analysis gave 
 
 Calculated for 
 2KI.TeI 4 .2H 2 0. 
 
 7.81 
 
 8.41 
 
 8.50 
 
 8.39 
 
 12.25 
 
 12.95 
 
 12.30 
 
 12.48 
 
 75.97 
 
 ... 
 
 76.68 
 
 76.11 
 
 3.57 
 
 ... 
 
 . 
 
 3.60 
 
POTASSIUM, RUBIDIUM, AND CAESIUM. 
 
 281 
 
 between 110-115, the resulting anhydrous salt being stable 
 at that temperature. This was shown by an iodine determina- 
 tion in the residue. Analysis gave 78.78 per cent of iodine ; 
 calculated for 2KI.TeI 4 , 78.94. 
 
 This salt is far more stable in the air than the corresponding 
 bromide, but the crystals lose their lustre in dry air, becoming 
 dull black on account of a superficial efflorescence. 
 
 Crystallography. 
 
 The crystallization of the anhydrous alkali-tellurium halides 
 is isometric. The chlorides were obtained in octahedra with 
 little or no modification, the bromides in combination of octa- 
 hedron and cube. The chlorides and bromides were measured, 
 and also proved to be isotropic by examination in polarized 
 light. Of the anhydrous iodides, the rubidium salt was the 
 only one obtained in crystals, and these were too small to 
 measure, but appeared under the microscope as combination of 
 octahedron and cube. They were so opaque that they could 
 not be tested in polarized light. 
 
 The two hydrous salts 2KBr.TeBr 4 .2H 2 O and 2KLTeI 4 . 
 2H 2 O, although analogous to each other in their composition, 
 differ in crystallization. The bromide is orthorhombic, as has 
 been shown by Baker,* also by Grailich and Lang in Rammels- 
 burg's Handbuch.* That the hydrous potassium tellurium 
 
 m 
 
 * Loc. cit. 
 
282 DOUBLE HALIDES OF TELLURIUM. 
 
 bromide obtained in the present work is identical with that 
 described by the above authors is shown by measurements of 
 the crystals. The crystallization of the salt 2KI.TeI 4 .2H 2 O is 
 monoclinic. Two different habits were observed ; long prisms, 
 developed in the direction of the clino axis were obtained from 
 hot solutions (Fig. 1). The mother-liquor from these, on 
 standing, gave shorter prisms in which the domes and clino- 
 pinacoids were wanting (Fig. 2). 
 The forms observed were : 
 
 m, 110 c, 001 d, 031 
 
 p, 111 b, 010 
 
 The axial ratio is as follows : 
 
 a : b : 6 0.7047 ; 1 ; 0.5688 B = 100 A 001 = 59 7' 16" 
 
 The crystals gave fair reflections of the signal on the goni- 
 ometer. Measurements chosen as fundamental are indicated 
 by an asterisk. 
 
 
 
 Calculated. 
 
 Measured. 
 
 Wl A Wl"' 
 
 , 110 A 1TO 
 
 62 20' 
 
 62 26' 
 
 m A bj 
 
 110 A 010 
 
 58 50' 
 
 *58 50' 
 
 b AC, 
 
 010 A 001 
 
 90 
 
 90 
 
 m A c } 
 
 110 A 001 
 
 63 57' 
 
 *63 57' 
 
 m A p, 
 
 110 A 11T 
 
 60 42' 
 
 *60 42' 
 
 c Ap, 
 
 001 A Til 
 
 55 21' 
 
 55 22' 
 
 b Ad, 
 
 010 A 031 
 
 34 20' 
 
 34 25' 
 
 b AP, 
 
 010 A 11T 
 
 61 42' 49" 
 
 61 42' 30" 
 
 m A w', 
 
 110 A T10 
 
 117 40' 
 
 117 33' 
 
 c A m f , 
 
 001 A T10 
 
 116 3' 
 
 116 11' 
 
 The crystals were too opaque for any optical examination. 
 
 In conclusion the author wishes to express his indebtedness 
 to Prof. H. L. Wells for valuable advice and for the interest 
 that he has taken in this work, and to Prof. S. L. Penfield 
 under whose direction the crystallography of these salts was 
 investigated. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 January, 1893. 
 
ON THE AMMONIUM-LEAD HALIDES.* 
 
 BY H. L. WELLS AND W. K. JOHNSTON. 
 
 IT is a well-known fact that the ammonium double halides, 
 like other ammonium salts, are usually analogous to those of 
 the alkali metals, especially to those of potassium. For ex- 
 ample, in connection with an investigation of the caesium- 
 mercuric halides, made by one of us, f it was noticeable that 
 caesium compounds were prepared, corresponding to all of the 
 four types of ammonium-mercuric halides that had been pre- 
 viously described. 
 
 A study of the caesium-lead and potassium-lead halides has 
 recently been made in this laboratory, J and, in view of the 
 fact that the existence of four very simple types of double 
 salts was established by that investigation, it seemed desirable 
 to re-investigate the ammonium-lead halides, since a consider- 
 able number of these with very complicated formulae had been 
 described. These peculiar ammonium-lead halides have formed 
 the most noticeable exception to the similarity of the alkali- 
 metal and ammonium salts, and it seems probable that it was 
 chiefly on account of these that Remsen has remarked : " The 
 position of the double halides containing ammonium is certainly 
 exceptional. They seem to be governed by some law of their 
 
 own." 
 
 All of the extremely complicated ammonium-lead salts have 
 been described by Andre*. || 
 
 The list of his alleged compounds is as follows : 
 
 * Amer. Jour. Sci., xlvi, July, 1893. 
 
 t Ibid., xliv, 221. I Ibid., xlv, 121. 
 
 Amer. Chem. Jour., xi, 296. 
 
 || Compt. Kend., xcvi, 435 and 1502 ; Bull. Soc. Chim., (2), xl, 14. 
 
284 ON THE AMMONIUM-LEAD HALIDES. 
 
 PbCl 2 .18NH 4 C1.4H 2 2PbBr 2 .14NH 4 Br.3H 2 
 
 PbCl 2 .10NH 4 Cl.H a O PbBr 2 .6NH 4 Br.H 2 
 
 2PbCl 2 .18NH 4 C1.3H 2 7PbBr 2 .12NH 4 Br.7H 2 
 
 PbCl 2 .6NH 4 Cl.H 2 3PbBr 2 .2NH 4 Br.H 2 
 
 4PbCl a .22NH 4 C1.7H 2 
 
 4PbCl 2 .18NH 4 C1.5H 2 
 
 4PbCl 2 .2KE 4 C1.6H 2 
 
 From Andre's original articles, it appeared that he made no 
 mention of having ever prepared any of his remarkable prod- 
 ucts more than once, and it seems probable that whenever he 
 obtained a crop of crystals or a precipitate he described it as 
 a new compound, without regarding the number of different 
 substances that it might contain. Andre* operated in two 
 ways. A part of his salts were made by dissolving a lead 
 halide in a hot solution of the corresponding ammonium 
 halide and cooling, while the rest were made by dissolving 
 lead monoxide in boiling solutions of ammonium chloride or 
 bromide. From the products made by the last method he 
 obtained lead oxychlorides or oxybromides by heating them 
 with water in sealed tubes. This peculiarity of these products 
 caused him to remark that all the chlorides made in this way 
 seemed to contain some oxychloride. This is the only evi- 
 dence of any suspicion on his part that he was obtaining mix- 
 tures. In a number of instances Andre* describes his products 
 as "crystalline precipitates," " brilliant plates with pearly 
 lustre," " crystallized bodies," etc., so that it would seem that 
 they should have been pure, but after having repeated his ex- 
 periments, following his methods as closely as his descriptions 
 permitted, with many variations and repetitions, we are con- 
 vinced that not one of the salts described by Andre exists. 
 
 Our work has resulted in the preparation of the following 
 series of salts : 
 
 Type 2 : 1. Type 1 : 1. Type 1 : 2. 
 
 3NH 4 PbCl 8 .H 2 O NH 4 Pb 2 Cl 6 
 
 (NH 4 ) 2 PbBr 4 .H 2 NH 4 Pb 8 Br 6 
 
 NH 4 PbI 8 .2H 2 O 
 
ON THE AMMONIUM-LEAD HALIDES. 285 
 
 For comparison, the potassium series, already referred to, is 
 given: 
 
 Type 2 : 1. Type 1 : 1. Type 1 : 2. 
 
 3KPbCl 8 .H 2 KPb 2 Cl 5 
 
 K 2 PbBr 4 .H 2 3KPbBr 8 .H 2 KPb 2 Br 5 
 
 KPbBr 8 .H 2 O .... 
 
 KPbI 3 .2H 2 .... 
 
 It is to be noticed that the two series correspond exactly, ex- 
 cept that no ammonium-lead bromide of the 1: 1 type was 
 obtained. These results show that the ammonium-lead halides 
 are entirely analogous to the potassium salts, and that there is 
 no indication that they are governed by a law of their own. 
 
 Returning to the mixtures described as compounds by An- 
 dre", it should be noticed that he came near finding the correct 
 formula) for three salts. His formula 4PbCl 2 2NH 4 C1.6H 2 O 
 would be correct if the water were omitted. He should 
 have found two more molecules of NH 4 Br in the formula 
 7PbBr 2 .12NH 4 Br.7H 2 O in order to have the salt (NH 4 ) 2 PbBr 4 
 H2O, and his formula 3PbBr 2 .2NH 4 Br.H 2 O, in comparison 
 with the errors involved in his more complicated mixtures, is 
 rather near NH 4 Pb 2 Br 6 . 
 
 Such work as Andrews is liable to hinder the development 
 of correct chemical theories. A case has been mentioned 
 above where it is probable that his results have been an im- 
 portant factor in causing the ammonium double halides to be 
 looked upon as a class of bodies distinct from the alkali-metal 
 compounds, and it may be mentioned that Carnegie * has used 
 Andrews formula PbBr 2 .6NH 4 Br in support of a theory on 
 double halides, although, as must be added, he considers 
 Andrews more complicated formulae as inconsistent with his 
 ideas. 
 
 We have investigated the ammonium-lead halides by meth- 
 ods which ( are entirely like those used for the potassium salts, 
 and, in addition, numerous experiments have been made in 
 order to investigate Andrews methods of preparation. The 
 
 * Amer. Chem. Jour., xv, 11. 
 
286 0^ THE AMMONIUM-LEAD HALIDES. 
 
 analytical methods used have been simple. Ammonium was 
 determined by distillation with potash solution and alkalime- 
 try. Lead was determined by treating the substance in a 
 platinum crucible with sulphuric acid, evaporating, igniting 
 and weighing lead sulphate. To determine a halogen, the 
 substance was treated with hot water and an excess of silver 
 nitrate was added ; after sufficient digestion, nitric acid was 
 added, and when the precipitate had settled properly it was 
 collected and weighed on a Gooch filter. Water was deter- 
 mined by the loss at 100, or sometimes at a somewhat higher 
 temperature. 
 
 1 : 1 Ammonium-Lead Chloride, SNHfl C1 3 .H 2 O. This 
 is formed by dissolving lead chloride in hot concentrated solu- 
 tions of ammonium chloride and cooling. Sample A was 
 made by dissolving 25 g. of PbCl 2 in 700 c. c. of an ammonium 
 chloride solution which was more than saturated when cold. 
 The double salt was deposited in the form of colorless, trans- 
 parent, prismatic crystals while the solution was still some- 
 what warm. Some of the crystals were removed from the 
 warm solution, quickly pressed between smooth filter-papers 
 and air-dried for analysis. On cooling the solution, ammo- 
 nium chloride crystallized upon the double salt. Sample B was 
 obtained by dissolving 5 g. of PbO in 200 c. c. of a boiling solu- 
 tion of NH 4 C1 which was nearly cold-saturated. The last 
 method was suggested by Andrews experiments. 
 
 Calculated for 
 x 3NH 4 PbCl s .H,0. 
 
 Ammonium . . 5.22- 5.45 . ..... 5.33 
 
 Lead .... 61.12-60.60 61.84-61.88 61.31 
 
 Chlorine . . . 31.78-31.79 31.64 31.56 
 
 Water .... 1.78 ...... 1.78 
 
 The water in A was determined by heating at about 120 
 for one hour. The limits of the conditions under which 
 this salt is formed are narrow, for, on slightly diluting the 
 solutions which have given it, the following compound is 
 produced. 
 
 1 : 2 Ammonium-Lead Chloride, Nff^Pb^Cl,. This salt 
 
ON THE AMMONIUM-LEAD HALIDES. 
 
 287 
 
 is produced under wide limits of conditions. It forms color- 
 less, short, transparent prisms which are usually doubly 
 terminated and are apparently orthorhombic in form. The 
 crystals retain their lustre on drying and are anhydrous. 
 Four crops were made under the following conditions : 
 
 A 
 B 
 C 
 D 
 
 NH 4 C1. 
 
 g- 
 
 100 
 100 
 200 
 200 
 
 PbCl 2 . 
 
 g- 
 
 30 
 
 20 
 15 
 60 
 
 Volume, 
 c. c. 
 
 1000 
 
 1000 
 
 550 
 
 700 
 
 These crops gave the following analyses : 
 
 A 
 
 Ammonium. 
 
 2.36-2.67 
 
 Lead. 
 
 67.38-67.36 
 
 Chlorine. 
 
 29.08-29.14 
 
 B . . . . 
 
 
 66 26-67.56 
 
 
 C 
 
 
 66.94-66.76 
 
 29.16-29.24 
 
 D 
 
 
 68.00-67.28 
 
 
 Calculated for | 
 NH 4 Pb 2 Cl 5 j 
 
 2.95 
 
 67.93 
 
 29.12 
 
 Another double chloride was observed, the composition of 
 which was not determined. It will be referred to beyond, 
 under the discussion of Andrews products. 
 
 2:1 Ammonium-Lead Bromide, (Nff 4 ) 2 Pbr 4 .H 2 0. This 
 salt is easily prepared by dissolving lead bromide in concen- 
 trated solutions of ammonium bromide. Its formation was 
 also observed when lead oxide was dissolved in ammonium 
 bromide by boiling. It forms beautiful radiating groups of 
 highly refracting, slender prisms. Three crops were made as 
 follows : 
 
 A 
 B 
 C 
 
 NH 4 Br. 
 
 200 
 ? 
 
 200 
 
 PbBr, 
 g. 
 
 50 
 25 
 
 50 
 
 Volume, 
 c. c. 
 
 380 
 260 
 380 
 
288 ON THE AMMONIUM-LEAD HALIDES. 
 
 The analyses were as follows : 
 
 Ammonium. Lead. Bromine. Water. 
 
 A ..... 6.01-5.86 37.12-36.84 55.06-55.10 2.60 
 
 B ........ 37.06-36.94 54.94 
 
 C ........ 37.26 ...... 
 
 65 - 08 3 - 10 
 
 1 : 2 Ammonium-Lead Bromide, Nff 4 Pb 2 jBr 5 . On slightly 
 diluting solutions from which the preceding salt would be 
 deposited, this salt is obtained. Repeated trials were made to 
 obtain a 1 : 1 salt intermediate between the two, but these 
 were without success. The salt forms square plates, often 
 several millimeters in diameter. The crystals darken some- 
 what on exposure to light, but they do not lose their lustre on 
 drying, and are evidently originally anhydrous. The com- 
 pound is formed under rather wide limits of conditions. A 
 single sample was analyzed. 
 
 ,, A Calculated for 
 
 Pound. NH 4 Pb 2 Br 6 . 
 
 Ammonium . . . 2.17- 2.17 2.16 
 
 Lead ..... 49.26-49.12 49.76 
 
 Bromine .... 48.28-48.22 48.08 
 
 1:1 Ammonium-Lead Iodide, Nff^Pbl^HzO. This was 
 the only double iodide obtained, although a thorough search 
 was made for other salts. It forms hair-like crystals of a pale 
 yellow color. Sample A was made by dissolving 100 g. of 
 NH 4 I and 10 g. of PbI 2 in sufficient hot water to make a volume 
 of 108 c. c., and cooling. Sample B was obtained by slightly 
 diluting the solution which gave A. It was noticed that 
 where lead iodide was deposited from a moderately concentrated 
 hot solution of ammonium iodide, the lead iodide disappeared 
 on cooling and in its place was formed a compact, silky mass of 
 crystals. Sample C was such a crop as just described, which 
 was carefully separated from the usual form of the double salt 
 which formed above in the solution. 
 
ON THE AMMONIUM-LEAD HALIDES. 289 
 
 Ammonium . 
 
 
 
 
 Calculated for 
 NH 4 PbI 3 .2H 2 O. 
 
 2.80 
 
 A. 
 
 . 2.40- 2.25 
 
 B. 
 
 2.97- 2.97 
 
 c. 
 
 Lead . . . 
 
 . 31.08-31.46 
 
 31.36-31.20 
 
 31.76 
 
 32.24 
 
 Iodine . . 
 
 . 59.76-59.70 
 
 59.85-59.75 
 
 62.45 
 
 59.36 
 
 Water 
 
 5.60 
 
 5.65 
 
 . 
 
 5.60 
 
 On Andre's Products. 
 
 Andre* prepared some of his most complicated products, such 
 as 2PbCl 4 .18NH 4 C1.3H 2 O and 4PbC1.22NH 4 C1.7H 2 O, by dis- 
 solving lead chloride in ammonium chloride solutions and by 
 the use of the corresponding bromides. On repeating his ex- 
 periments with the bromides, we have not observed anything 
 which did not correspond to the salts which we have described, 
 or to mixtures of these with ammonium bromide. He de- 
 scribes his product 4PbCl 2 . 22NH 4 C1.7H 2 O as an abundant pre- 
 cipitate of very brilliant plates with a pearly lustre. We have 
 often observed this beautiful precipitate, but, after many 
 attempts, we have been unable to determine its composition 
 with certainty. It apparently forms only in warm solutions 
 which are almost saturated with ammonium chloride. The 
 prismatic salt NH 4 Pb 2 Cl 5 is usually deposited just before the 
 plates begin to form, and it often forms with them. Large 
 amounts of ammonium chloride often crystallized out when 
 attempts were made to separate the precipitate from the mother- 
 liquor. This happened often even when the solution with the 
 suspended precipitate was poured upon a large mass of filter- 
 paper, so as to soak up the liquid as quickly as possible. Such 
 a crop, which was granular and showed little evidence of being 
 composed of plates, gave 30.50 per cent of lead. By collect- 
 ing the precipitates, in the manner just described, at an earlier 
 period in the process of their formation, we succeeded in 
 obtaining crops that were apparently pure, but the plates were 
 so exceedingly thin and small, and the mother-liquor was so 
 concentrated, that we had little confidence in the purity of 
 these products. Two such crops gave 51.13 and 53.69 per 
 cent of lead. The formula (NH 4 ) 2 PbCl 4 requires 53.77, and 
 it is possible that this may be the true f ormula for the sub- 
 
 19 
 
290 ON THE AMMONIUM-LEAD HALIDES. 
 
 stance. It is certain that these two products were practically 
 free from the prismatic salt NH 4 Pb 2 Cl 6 , so that it can be posi- 
 tively stated that the ratio of ammonium to lead cannot be 
 greater than 2 : 1 in this compound. Andrews formula is, 
 therefore, very far from correct. He must have analyzed a 
 mixture of the plates with ammonium chloride. We have 
 found all such mixtures to be anhydrous after being dried in 
 the air for a short time, hence the water in Andrews formula is 
 remarkable. 
 
 A number of attempts were made to obtain the compound 
 in a purer state by warming the solutions and by diluting them 
 slightly, after the precipitates had formed, but these operations 
 left the products open to suspicion, since it was found that 
 further dilution completely decomposed the plates. Several 
 crops, made in this way, gave 59.01, 57.64, 53.69, and 57.80 
 per cent of lead, which is an insufficient amount for the 1 : 1 
 anhydrous salt, requiring 62.4. This may be its composition, 
 but it is possible also that it is a dimorphous form of 
 NH 4 Pb 2 Cl 5 , for the plates, in being square with diagonal mark- 
 ings, resemble the salt KPb 2 Br 6 and other bromides of this type. 
 The undetermined double chloride can be readily prepared by 
 dissolving 160 g. of ammonium chloride and 25 g. of lead chlo- 
 ride in sufficient boiling water to make a volume of 400 c. c. 
 and allowing the solution to cool slowly. The compound 
 often forms in such abundance as to completely fill the solution 
 with a loose mass of the very thin plates. 
 
 Andrews other method of preparing his products was by dis- 
 solving lead monoxide in ammonium chloride and bromide 
 solutions. He describes only one complicated bromide, PbBr 2 . 
 6NH 4 Br.H 2 O, made in this way. We have made a number of 
 experiments with ammonium bromide and lead oxide without 
 obtaining anything but our own salts and mixtures. Since a 
 number of his chlorides were made by dissolving lead oxide in 
 ammonium chloride solutions, we have made a very careful 
 study of the products obtained by this operation. Andre* 
 sometimes indicates the length of time of boiling the ammo- 
 nium chloride solution with lead oxide, but is uncertain how 
 
ON THE AMMONIUM-LEAD HALTDES. 291 
 
 rapidly he boiled his solutions. We have therefore made 
 numerous experiments with wide variations in the amount of 
 ammonia boiled off. Among the various products that we 
 obtained, including the salts that we have described, we fre- 
 quently noticed a substance that appeared to be new. It was 
 deposited after the solutions had become cold or nearly so, 
 forming brilliant crystals, apparently nearly cubic in form, but 
 so much rounded as to have no distinct faces. These crystals 
 were often 1 or 2 mm. in diameter. They formed upon the bot- 
 tom and sides of the beaker, adhering firmly to the glass, and 
 their quantity was sometimes such that the walls of the vessel 
 were thickly studded with them. This product evidently 
 corresponded with one of Andrews, for he says of " PbCl 2 . 
 18NH 4 C1.4H 2 O " that it is a very hard crystalline deposit 
 adhering to the glass. We encountered considerable difficulties 
 in obtaining the substance in a pure condition, for it was in- 
 clined to deposit upon other things that had previously formed, 
 and it adhered to them as well as to the glass. The first crop 
 analyzed (No. 1) was evidently not pure. Two other crops 
 (2 and 3) appeared better, but still not quite pure. At last, by 
 decanting a solution just before these crystals began to form, 
 we obtained a crop (No. 4), that seemed entirely satisfactory. 
 The following analyses of the four crops were made: 
 
 Chlorine. 
 
 58.05 = 99.77 
 
 No. 1 
 No. 2 
 
 Ammonium. 
 
 . . . 28.28 
 
 Lead. 
 
 13.44 
 6.51 
 
 No. 3 
 
 No. 4 
 
 
 
 6.74 
 1.08 
 
 
 The crystals of the last product had exactly the same appear- 
 ance as the others. It is evident that lead is not an essential 
 constituent of the substance, and that the substance is am- 
 monium chloride crystallized in an unusual form. Analysis 
 No. 1 corresponds to PbCl 2 .24NH 4 Cl. This is not far from 
 Andrews formula, and it shows what he probably analyzed. It 
 is to be noticed that, while our impure ammonium chloride 
 was practically anhydrous, Andre* gives a considerable amount 
 
292 ON THE AMMONIUM-LEAD HAL1DES. 
 
 of water in his formula. It seems probable that he did not 
 properly dry his products before analyzing them, and, more- 
 over, he evidently determined water from the deficiencies in 
 his analyses. There is more or less water in every one of his 
 formulae. 
 
 Since it was evident that Andrews operation of boiling am- 
 monium chloride solutions with lead oxide could be imitated, 
 with possibilities for greater variations, by adding ammonia to 
 solutions of lead chloride in ammonium chloride, we have car- 
 ried out a series of experiments on that plan. No indications 
 of the existence of any of Andrews complicated compounds 
 were obtained in this way, but, besides the form of ammonium 
 chloride that adheres to the glass, a peculiar modification of it 
 in the form of large, transparent pointed crystals with no dis- 
 tinct faces was observed. A sample of these contained 3.23 
 per cent of lead. When much ammonia was used in these 
 experiments, an oxychloride of lead was obtained. A pure 
 product of this was prepared by saturating a cold-saturated 
 solution of ammonium chloride with lead chloride while boiling, 
 then adding an equal volume of the cold-saturated ammonium 
 chloride solution and finally adding a large excess of ammonia. 
 A precipitate was formed and re-dissolved by the ammonia. 
 The oxychloride was deposited, on cooling, in the form of small, 
 blade-like, transparent crystals. Analysis showed that it was 
 the compound PbClOH. 
 
 Calculated for 
 PbClOH. 
 
 Lead ...... 79.17 79.77 
 
 Chlorine .... 14.06 13.68 
 
 Oxygen ..... [2.72] 3.08 
 
 Water ..... 4.05 3.47 
 
 This compound has long been well known, having been used 
 as a white pigment. It is worthy of remark that Andre* re- 
 described this body correctly in one of his papers that has 
 already been referred to.* He made it by heating a small 
 quantity of " PbCl 2 NH 4 Cl.H 2 O " with water in a sealed tube. 
 
 * Bull. Soc. Chim., II, xl, 15 (1883). 
 
ON THE AMMONIUM-LEAD HALIDES. 
 
 293 
 
 It is evident that such a product could not have been produced 
 if the chlorides had not contained some basic substance. 
 
 Experiments with Ammonium Chloride and Lead Iodide. 
 
 Poggiale * and Volkel f have each described a mixed double 
 halide of ammonium and lead, neither of which agrees with 
 the types of unmixed halides which we have described in this 
 article. Poggiale's formula is PbI 2 .4NH 4 C1.2H 2 O, while Vol- 
 kel's is PbI 2 .3NH 4 Cl. It is evident that the two investigators 
 obtained the same compound, for both made their products in 
 essentially the same way, and both describe them as occurring 
 in the form of silky needles. We have repeated the experi- 
 ment of dissolving lead iodide in ammonium chloride solutions 
 and have readily obtained the silky crystals. The product 
 resembles exactly the salt NH 4 PbI 8 .2H 2 O, except that it is 
 considerably paler in color than the latter. An analysis of a 
 pure product, carefully dried by pressing with paper, showed 
 that the formulse of both Poggiale and Volkel are incorrect, 
 and that their products must have been contaminated with 
 ammonium chloride. The salt corresponds in type to our 
 double iodide. It loses one molecule of water on exposure to 
 the air for two or three days, becoming much darker in color ; 
 the second molecule of water goes off at 100 . 
 
 Found. 
 
 Ammonium . . . 3.33 
 
 Lead 34.83-34.68 
 
 Chlorine .... 5.11- 5.10 
 
 Iodine 51.08-50.94 
 
 Water . 5.35 
 
 Calculated for 
 NH 4 Cl.PbI 2 .2H,0. 
 
 3.27 
 37.60 
 
 6.45 
 46.14 
 
 6.54 
 
 The analysis indicates that the lead iodide and ammonium 
 chloride do not combine quite unchanged, and that the formula 
 NH 4 Pb(Cl,I) 3 .2H 2 O possibly expresses the composition of the 
 product better than the one given above. In the course of our 
 experiments with ammonium chloride and lead iodide, we 
 obtained a crop of crystals which showed an almost complete 
 
 * Compt. Rend., xx, 1180. 
 
 t Pogg. Ann., Ixii, 252. 
 
294 ON THE AMMONIUM-LEAD HALIDES. 
 
 replacement of the iodine by chlorine. The salt was a 1 : 2 
 compound, a type which apparently does not exist among the 
 double iodides of lead. It gave the following analysis : 
 
 Found. 
 
 Calculated for 
 
 NH 4 Pb 2 Cl 5 . 
 
 Ammonium 2.95 
 
 Lead 66.82 67.93 
 
 Chlorine 28.76 29.12 
 
 Iodine 1.44 . . . 
 
 The view that certain mixed lead double halides are to be 
 considered as variable mixtures of two isomorphous, unmixed 
 double halides was arrived at by Wells and Wheeler * from 
 experiments with caesium chloride and lead bromide. Herty f 
 has more recently arrived at the same conclusion from his 
 investigation of the mixed double bromide and iodide of potas- 
 sium and lead. It is not safe to conclude, however, because 
 definite mixed double halides do not occur in some cases, that 
 they are never formed. It is certain that there is a tendency 
 toward the formation of such definite compounds in many cases. 
 Tor example, it is to be noticed that in the above analysis of 
 NH 4 Cl.PbI 2 .2H 2 O there is only a slight variation from the 
 composition required by the formula, while, from a solution of 
 the same ingredients, another type of salt was deposited which 
 was almost free from iodine. The compound Cs 2 HgCl 2 I 2 , J 
 described by one of us, is evidently a definite mixed double 
 halide which is not intermediate in its properties between the 
 corresponding chloride and iodide, and which has a constant 
 composition. A number of other caesium-mercuric halides 
 were described, which approached a constant composition when 
 made under varying conditions. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 March, 1893. 
 
 * Amer. Jour. Sci., Ixr, 129. 
 t Amer. Chem. Jour., xv, 81. 
 I Amer. Jour. Sci., xliv, 232. 
 
ON THE RUBIDIUM-LEAD HALIDES, AND A 
 
 SUMMARY OF THE DOUBLE HALIDES 
 
 OF LEAD.* 
 
 BY H. L. WELLS. 
 
 THE caesium-lead and potassium-lead halides have already 
 been studied in this laboratory, and an account of the ammo- 
 nium compounds is given in the preceding article. It has 
 therefore seemed desirable to make an investigation of the 
 rubidium salts in order to make the work more complete. 
 
 2 : 1 Rubidium-Lead Chloride, 2Rb z PbCh.E z O. This was 
 formed by dissolving lead chloride in a solution of rubidium 
 chloride which was so concentrated as to be almost saturated 
 when cold. It forms colorless, transparent, slender, flat prisms 
 which retain their lustre on exposure to the air. Two separate 
 crops were analyzed. 
 
 rt Calculated for 
 
 Pound. 2Kb t PbCl 4 .H,0. 
 
 Rubidium . . . 29.63 29.85 32.39 
 
 Lead 41.41 41.75 39.20 
 
 Chlorine. . . . 26.73 26.84 26.90 
 
 Water .... 2.29 2.01 1.51 
 
 100.06 100.45 100.00 
 
 The amount of water in the salt seems somewhat uncertain, 
 but, since there was no evidence of loss of water by efflores- 
 cence and since the salt was simply air-dried without being 
 pulverized, the above formula is preferred to Rb 2 PbCl 4 .H 2 O, 
 which requires 2.99 per cent of water. The water was deter- 
 mined by heating to about 200 ; at 100 the salt lost only 
 about one quarter of its water in twelve hours. 
 
 1:2 Rubidium-Lead Chloride, RbPb^Cl^. This com- 
 pound forms small, prismatic crystals which are usually 
 
 * Amer. Jour. Sci., xlvi, July, 1893. 
 
296 ON THE RUBIDIUM-LEAD HALIDES. 
 
 grouped side by side in nearly parallel position. It is pro- 
 duced from solutions which are more dilute than those from 
 which the preceding salt is prepared, and it is formed under 
 rather wide limits of conditions. Two separate crops gave 
 the f oUowing analyses : Calculated for 
 
 Found - EbPb 2 Cl 6 . 
 
 Rubidium . . . 13.09 12.68 12.63 
 
 Lead 60.57 61.05 61.15 
 
 Chlorine . . . 26.19 26.29 26.22 
 
 99.85 100.02 100.00 
 
 2 : 1 Rubidium-Lead Bromide, ^Rb 2 PbBr 4 .H 2 0. This 
 salt resembles the corresponding chloride, both in its formation 
 and appearance. Two crops gave the following analyses : 
 
 _ , Calculated for 
 
 Found - 2Rb 2 PbBr 4 .H 2 0. 
 
 Kubidium . . . 23.17 22.73 24.19 
 
 Lead .... 30.29 30.81 29.29 
 
 Chlorine 45.04 45.26 
 
 Water .... 1.55 1.51 1.27 
 
 100.09 100.00 
 
 1 : 2 Rubidium-Lead Bromide, RbPb 2 Br 5 . This forms 
 square plates. It is readily prepared, since it is formed under 
 considerable variations of conditions. 
 
 w , Calculated for 
 
 Found - RbPb 2 Br 6 . 
 
 Rubidium . . . 9.81 9.50 
 
 Lead 45.74 46.03 
 
 Bromine .... 44.62 44.47 
 
 100.17 100.00 
 
 111 Rubidium-Lead Iodide, RbPbI s .2H 2 0. This is the 
 only double iodide that could be produced under widely vary- 
 ing conditions. It forms very slender, hair-like prisms of a 
 pale yellow color. It rapidly loses its water when exposed to 
 the air, undergoing a remarkable change of color. The pale 
 yellow compound quickly assumes an orange color, then the 
 color becomes almost like that of the original salt. It is evi- 
 dent that the salt, which contains two molecules of water, 
 loses a part of this, probably one molecule, with change of 
 color to orange ; then the remainder of the water is lost with 
 
ON THE RUBIDIUM-LEAD HALIDES. 297 
 
 another change of color. It is interesting to notice, in this 
 connection, that the salt NH 4 PbClI 2 .2H 2 O undergoes a similar 
 change of color, as far as the first step is concerned, on losing 
 one molecule of water when it is exposed to the air, but this 
 salt does not lose its second molecule at ordinary temperatures. 
 A sample of the rubidium-lead iodide was rapidly pressed on 
 paper until some of the particles began to show a change of 
 color to orange. Water was determined in this sample from 
 the loss at 100. 
 
 RVmnH Calculated for 
 
 Found. RbPbI 8 .2H 2 0. 
 
 Water 6.09 5.07 
 
 An air-dry sample gave the following analysis : 
 
 Pound. Calculated. 
 
 Kubidium . . . 13.29 12.70 
 
 Lead 28.95 30.73 
 
 Iodine .... 56.80 56.57 
 
 99.04 100.00 
 
 Summary. 
 
 The following table gives a list of the lead double halides 
 which have been prepared in this laboratory. All of them 
 were new compounds except KPbBr 8 .H 2 O and KPbI 8 .2H 2 O, 
 these having been previously described by Remsen and Herty. 
 
 4:1 2:1 1:1 1:2 
 
 Cs 4 PbCl 6 CsPbClg CsPb 2 Cl 6 
 
 Cs 4 PbBr 6 CsPbBr 8 CsPb 2 Br 6 
 
 CsPbI 3 
 
 2Rb 2 PbCl 4 .H 2 RbPb 2 Cl 6 
 
 2Rb 2 PbBr 4 .H 2 O RbPb 2 Br 6 
 
 RbPbI 8 .2H 2 
 
 3KPbCl 8 .H 2 KPb 2 Cl 6 
 
 K 2 PbBr 4 .I! 2 3KPbBr 8 .H 2 O KPb 2 Br B 
 
 KPbBr 8 .H 2 O 
 
 KPbI 8 .2H 2 O 
 
 3NH 4 PbCl 8 .H 2 NH 4 Pb 2 Cl 5 
 
 (NH 4 ) 2 PbBr 4 .H 2 NH 4 Pb 2 Br 6 
 
 NH 4 PbI 8 .2H 2 
 
 NH 4 PbClI a .?H 8 
 
298 ON THE RUBIDIUM-LEAD HALIDES. 
 
 An inspection of the table shows that the caesium salts differ 
 from the others in including the 4 : 1 type and in being with- 
 out any 2 : 1 salts. It is quite probable that we have not 
 succeeded in preparing all the salts that are possible, but it 
 seems certain that the 4:1 rubidium, potassium, and ammo- 
 nium salts cannot be made on account of the comparative in- 
 solubility of the simple halides. The caesium salts also differ 
 from the others in being all anhydrous. The hydrous rubidium 
 salts have less water or lose it more readily than the potassium 
 compounds. KPbI 8 .2H 2 O is stable in the air, but RbPbI 8 .2H 2 O 
 loses its water readily. There is evidently a gradation, in af- 
 finity for water, from the caesium to the potassium compounds. 
 A gradation in water, from the chloride to the iodide, appar- 
 ently exists in the potassium and ammonium compounds of 
 the 1 : 1 type. That such gradations in water exist among 
 the double halides, increasing with the atomic weight of the 
 halogens and decreasing with the atomic weight of the alkali 
 metals, has already been observed by Remsen.* 
 
 The simplicity of the ratios in the four types of double hal- 
 ides of lead is noticeable. The 4 : 1 type, according to 
 Werner's remarkable theory, f may be considered as the ideal 
 type of a double halide of an alkali metal and a bivalent 
 metal, and as the limit beyond which the ratios of alkali metal 
 to lead cannot go. The type is represented, as Werner 
 mentions, by numerous double cyanides of bivalent metals, 
 such as K 4 Fe(CN) 6 and by other salts, such as K 4 CdCl 6 . 
 
 The number of 2 : 1 lead salts that we have prepared is 
 rather small, but this is a very common type among the known 
 double halides of the other bivalent metals. 
 
 The number of 1 : 1 lead salts is the largest of all. It is 
 remarkable that all the double iodides belong to this type. 
 This is also a well-known type of bivalent metal double hal- 
 ides. It is noticeable that the salt CsPbBr 3 is dimorphous, 
 while three mercuric salts of the same type, J CsHgCl 8 , 
 CsHgBr 8 , and CsHgClBr 2 are also dimorphous. 
 
 * Amer. Chem. Jour., xir, 88. f Zeitschr. anorg. Chem., iii, 281. 
 
 \ Amer. Jour. Sci., xliy, 222. 
 
ON THE RUBIDIUM-LEAD HALIDES. 299 
 
 The 1 : 2 salts are all anhydrous, and a chloride and bro- 
 mide were prepared with each of the alkali metals and with 
 ammonium. They are formed under wide limits of condi- 
 tions and are therefore very easily prepared. It is noticeable 
 that the rubidium, potassium, and ammonium chlorides of this 
 type all crystallize in prisms, while all the bromides and the 
 chloride containing cassium crystallize in plates. A number 
 of other double halides of this type are known, especially 
 among the mercuric compounds. Herty has evidently pre- 
 pared a potassium-lead double halide of the 1 : 2 type contain- 
 ing bromine and iodine, although he interprets his results in 
 an entirely different way. In a recent article * he describes 
 some tabular crystals of an olive-green color which he has 
 selected and analyzed with evident care and skill. He gives 
 the following analyses of three separate products : 
 
 p. 95, D . 
 p. 95, E . 
 p. 104 . 
 
 The following ratios may be derived from the above 
 analyses : 
 
 Pb : I + Br : K. 
 
 p. 95, D . . . 2. 5.01 1.07 
 
 p. 95, E ... 2. 5.15 1.08 
 
 p. 104 .... 2. 5.08 1.08 
 
 The ratio required for the formula KPb 2 (Br,I) 6 is Pb: 1 + 
 Br : K = 2 : 5 : 1, and the agreement is so close that there can 
 be no doubt that this is the formula. Although no pure 
 iodide of this type has been produced, it is interesting to 
 notice that Herty's compound shows that the potassium salt 
 is capable of existence when mixed with a relatively large 
 amount of the bromide. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 March, 1893. 
 
 * Amer. Chem. Jour., xv, pp. 94, 95, 97-99, 103, 104 (February, 1893). 
 
 Pb. 
 
 i. 
 
 Br. 
 
 K. 
 
 44.89 
 
 17.61 
 
 32.59 
 
 4.56 = 99.65 
 
 45.42 
 
 14.94 
 
 34.72 
 
 4.50 = 99.58 
 
 43.87 
 
 22.20 
 
 29.03 
 
 4.43 = 99.53 
 
ON THE DOUBLE HALIDES OF ARSENIC WITH 
 CJESIUM AND KUBIDIUM; AND ON SOME COM- 
 POUNDS OF AESENIOUS OXIDE WITH THE HAL- 
 IDES OF CAESIUM, RUBIDIUM, AND POTASSIUM.* 
 
 BY H. L. WHEELER. 
 
 No compounds of arsenious halides with alkaline halides 
 have been definitely described. Nickldsf, in his work on the 
 bromides and iodides of arsenic, antimony, and bismuth, states 
 that these salts combine with alkaline bromides and iodides 
 respectively, but in the case of arsenic he gives no analyses of 
 the compounds which he obtained, and he does not describe 
 the methods that he used in preparing them. He does not 
 even state with what alkaline halides he performed his experi- 
 ments. Emmet,J Harms, Schiff, and Sestini|| have described 
 compounds of arsenious oxide with potassium halides, but this 
 class of bodies has been most thoroughly studied by Riidorff.lT 
 His results indicate the existence of two types of this class of 
 compounds, the first containing one molecule of alkaline 
 halide to one of arsenious oxide, the other having these 
 constituents in the ratio 1:2. In the present investigation a 
 complete series of the caesium and rubidium oxyhalides of the 
 1 : 1 type was obtained, while the formation of the 1 : 2 type 
 was not observed. There is evidently a gradation in stability 
 from the oxychlorides to the oxyiodides, the stability increasing 
 with increasing atomic weight of the halogens. Attempts to 
 prepare double halides of pentavalent arsenic were without 
 success. 
 
 * Amer. Jour. Sci., xlvi, August, 1893. 
 
 t Compt. Rend., xlviii, 839 ; Jour. Pharra., Ill, xli, 142 ; Rep. chim. pure, 
 i,366. 
 
 t Amer. Jour. Sci., I, xviii, 58. Ann. Chem. Pharm., xci, 371. 
 
 II Ibid., 228, 72. 
 
 T Ber. d. Deutsch. Chem. Ges., xix. 2668 ; Ibid., xxi, 3053. 
 
DOUBLE HALIDES OF ARSENIC. 301 
 
 The new compounds to be described in this article are the 
 following : 
 
 3CsC1.2AsCl 8 3CsBr.2AsBr 8 3CsI.2AsI 8 
 
 3BbC1.2AsCl, 3EbBr.2AsBr 8 3RbI.2AsI 8 
 
 CsCl.As 2 3 CsBr.As 2 O 8 CsLAs 2 8 
 
 EbCl.As a 8 KbBr.As 2 8 RbI.As 2 8 
 
 The compound As 2 O 8 .KI, which had not been observed by 
 Riidorff,* was obtained in attempts to prepare a double iodide 
 of potassium and arsenic. 
 
 It is to be noticed that only one type of double halides was 
 obtained. This type corresponds to the most readily prepared 
 double chloride of caesium with antimony, described by Rem- 
 sen and Saunders f and of caesium and bismuth, described by 
 Remsen and Brigham.J Many attempts were made to prepare 
 arsenic double halides of other types than this single one, but 
 these have invariably been without success, although several 
 types of antimony and bismuth double halides have been 
 described. 
 
 It has been shown in several instances by Wells and Wheeler 
 that csesium and rubidium halides form more stable or more 
 complete series of double salts than the halides of the other 
 alkali metals. This fact is again well illustrated in the double 
 halides of arsenic, for the caesium and rubidium double halides 
 are prepared without difficulty, while with potassium no double 
 halides were obtained. 
 
 Methods of Preparation. 
 
 To prepare the double halides, a strong acid solution is 
 necessary, in order to prevent the decomposition of the arseni- 
 ous halide and the consequent formation of oxy-compounds. 
 The double halides are less soluble in the strong than in the 
 dilute halogen acids. Excess of one or the other of the con- 
 stituents has no effect on the composition of the products 
 
 * Loc. cit. t Amer. Chem. Jour., xiv, 152. 
 
 t Ibid., p. 164. Amer. Jour. Sci., Ill, xliii, 475 ; III, xliv, 42. 
 
302 DOUBLE HALIDES OF ARSENIC 
 
 obtained. The formation of As 2 O 8 compounds was observed 
 on treating the double halides or a solution of the constituents 
 in strong acids, with water or dilute acids. The oxygen com- 
 pounds are difficultly soluble in dilute acids; strong acids 
 convert the caesium and rubidium compounds into the double 
 halides. 
 
 Method of Analysis. 
 
 The salts were filtered on the pump, and without delay were 
 carefully freed from the mother-liquor by pressing on paper. 
 They were then dried in the air. In no case was water used 
 to wash them. In order to determine arsenic the salt was dis- 
 solved in the cold in HC1 sp. gr. 1.1, and hydrogen sulphide 
 passed in for about one hour ; then a little alcohol was added, 
 the whole was warmed on the water bath for a short time, in 
 order to drive off the excess of hydrogen sulphide and effect 
 the separation of the last traces of arsenious sulphide. The 
 sulphide of arsenic was collected on a Gooch filter, and after 
 washing with water, alcohol, and carbon disulphide it was dried 
 at 100 and weighed. Sulphuric acid was added to the filtrate, 
 and the alkali metal determined as normal sulphate by evapo- 
 rating and igniting the residue in a stream of air containing 
 ammonia. The halogens were determined in a separate por- 
 tion as silver halides in the usual manner. 
 
 Caesium and Rubidium Arsenious Chlorides, SCsCL^AsOl^ 
 and 3KbCl.2AsCl z . These have a pale yellow color, like the 
 corresponding antimony and bismuth double chlorides. The 
 caesium salt was obtained by dissolving 250 g. of CsCl in 
 dilute HC1. 2 g. of As 2 O 8 in dilute HC1 were then added. 
 This produced a precipitate which dissolved on the addition 
 of about 2 liters of hot HC1 sp. gr., 1.1. On cooling, light 
 yellow crystals were deposited. A portion of these was 
 recrystallized from a strong HC1 solution of AsCl 8 . The 
 rubidium salt was prepared in the same manner, except that 
 much stronger solutions were required. Saturated solutions 
 of rubidium and arsenic chlorides in 20 per cent HC1 produce 
 no precipitate on mixing, but if concentrated HC1 is added. 
 
WITH CAESIUM AND RUBIDIUM. 
 
 303 
 
 brilliant spangles of the double salt separate. The analysis of 
 these products gave : 
 
 Prepared with Ex- 
 cess of CsCl. 
 
 Cs 46.14 
 
 As 17.15 
 
 Cl 36.89 
 
 45.27 
 
 36.74 
 
 Found. 
 
 Prepared with Ex- 
 cess of AsCl 3 . 
 
 45.09 
 17.11 
 36.12 
 
 Calculated for 
 3CsC1.2AsCl s . 
 
 45.94 
 17.27 
 36.79 
 
 Kb 35.55 
 
 As ..... 20.14 
 Cl 44.04 
 
 Calculated for 
 3RbC1.2AsCl 8 . 
 
 35.33 
 20.66 
 44.01 
 
 Both salts can be recrystallized from hydrochloric acid of 
 sp. gr. 1.1. 100 parts of HC1 sp. gr. 1.2 dissolve .429 parts of 
 the caesium salt and 2.935 parts of the rubidium compound. 
 Since the corresponding potassium salt apparently does not 
 exist, these solubilities suggest a convenient method for ob- 
 taining caesium and rubidium free from potassium. 
 
 Ccesium and Rubidium Arsenious Bromides, 3CsBr.2AsBr^ 
 and 3KbBr.AsBr y These are amber-yellow, the shade 
 being somewhat darker than that of the chlorides. They 
 are most conveniently prepared by using an excess of the 
 alkaline halide. Strong hot solutions of the alkaline bro- 
 mides were made in about 40 per cent HBr. On adding 
 crystals of AsBr 3 these melted, but soon solidified to a yellow 
 mass of the double halide. This dissolved on boiling, and, on 
 cooling, brilliant yellow crystals were obtained. These com- 
 pounds can be recrystallized unaltered from strong HBr. 
 Analysis gave : 
 
 Cs . . . . 
 
 Pound. 
 
 . . . 31.91 
 
 Calculated for 
 3CsBr.2AsBr 8 . 
 
 31.44 
 
 As . . . . 
 
 . . . 11.89 
 
 11.82 
 
 Br 
 
 56.94 
 
 56.74 
 
 Prepared with Ex- Prepared with Ex- 
 cess of RbBr. cess of AsBr 3 . 
 
 Kb 
 As 
 Br 
 
 23.35 
 12.55 
 63.97 
 
 64.43 
 
 Calculated for 
 3RbBr.2AsBr,. 
 
 22.77 
 13.31 
 63.92 
 
304 DOUBLE HALIDES OF ARSENIC 
 
 Ccesium and Rubidium Arsenious Iodides, 3CsI.2AsI z and 
 3RbI.2AsI y These are deep red, the larger crystals of the 
 csesium compound are more opaque and appear black. To 
 prepare these compounds the normal alkaline iodides were dis- 
 solved in strong colorless hydriodic acid, and these solutions 
 were then saturated boiling with crystals of Asia. Unless the 
 hydriodic acid is decolorized the product obtained in the case 
 of the csesium salt is generally impure, being mixed with 
 CsI 3 .* 
 
 A well-crystallized product of the double csesium salt was 
 obtained by preparing the salt in the presence of considerable 
 alcohol. Analysis of these compounds gave 
 
 ,, , Calculated for ip nv , n A Calculated for 
 
 Found. 3C8 I.2A8l 8 . 3RbI.2AsI 8 . 
 
 Cs . . 24.38 23.58 Kb . . 16.86 . . . 16.55 
 
 As . . 8.92 8.87 As . . 9.96 10.60 9.68 
 
 I ... 67.23 67.55 I ... 73.65 . . . 73.77 
 
 An attempt was made to prepare potassium arsenious chlo- 
 ride by mixing solutions of potassium chloride and arsenious 
 acid, saturated solutions of these substances in concentrated 
 HC1 being used for the purpose. No precipitate was thus 
 produced, and, on concentrating the solution, potassium chlo- 
 ride was deposited. Aqueous solutions of potassium chloride, 
 when added to solutions of arsenic trioxide in concentrated 
 HC1, gave precipitates consisting chiefly of As 2 O 8 . Analogous 
 experiments with potassium bromide and arsenious bromide 
 gave similar results, and operating in the same way with KI 
 and AsI 3 in concentrated HI solutions, nothing but crystals of 
 AsI 8 or mixed crops of AsI 3 and KI were obtained. Similar 
 negative results, in respect to the formation of double halides 
 of ammonium and arsenic, have been obtained by Wallace, f 
 
 Compounds of Arsenic Trioxide with Alkaline Halides^ 
 Cs Cl.As 2 3 and Mb Cl.As z O s . When a hot saturated aqueous 
 solution of 25 g. of csesium chloride was saturated with 
 3CsC1.2AsCl 8 , a finely divided white precipitate was formed on 
 
 * Am. Jour. Sci., xliii, 17 ; Zeitschr. anorg. Chem., i, 85. 
 t Phil. Mag., IV, xvi, 358; xvii, 122, 261. 
 
WITH CAESIUM AND RUBIDIUM. 305 
 
 cooling (analysis 1). When 6.5 g. of the double chloride 
 were dissolved in 800 c. c. of a cold-saturated solution of As 8 O 3 
 in HC1 sp. gr. 1.1 by the aid of heat, a similar precipitate 
 was obtained (analysis 2). Products intermediate in composi- 
 tion were obtained by recrystallizing the double halide from 
 water (analysis 3), from 10 per cent HC1 (analysis 4), and 
 from 15 per cent HC1 (analysis 5). 
 
 Cs . . . 41.51 41.85 41.75 41.80 25.56 34.29 
 
 As ... 35.59 34.58 35.37 . . . 47.64 43.05 
 
 Cl . . . . 11.46 . . . 11.31 . . . 11.64 8.86 
 
 O . . . . [11.44] . . . [11.57] . . . [15.16] [13.80] 
 
 Cs 
 
 4. 
 
 35.30 
 
 5. 
 
 33.92 
 
 uaicmaiea 101 
 CsCl.As 2 O s . 
 
 36.29 
 
 As 
 
 39.14 
 
 38.10 
 
 40.93 
 
 Cl 
 
 10.63 
 
 11.42 
 
 9.68 
 
 
 
 [14.93] 
 
 [16.56] 
 
 [13.10] 
 
 The analyses show a considerable variation from the com- 
 position required by the formula, the products made under the 
 extreme conditions giving ratios of arsenious oxide to caesium 
 chloride of 3 : 4 and 3 : 2 instead of 1:1. The conditions 
 varied so widely, however, that it seems fair to assume the 
 existence of a 1 : 1 compound. 
 
 When the double chloride of rubidium was recrystallized 
 from about 15 per cent HC1, a white crystalline crust was 
 obtained on slowly cooling. This gave analytical results 
 agreeing with the formula RbCl.As 2 O 8 . 
 
 Calculated for 
 
 Rb ...... 26.90 26.80 
 
 As .......... 47.03 
 
 Cl ....... 11.41 11.13 
 
 O .......... 15.04 
 
 Under the microscope these compounds appear as irregular 
 grains or plates, of indefinite crystalline form. 
 
306 
 
 DOUBLE HALIDES OF ARSENIC 
 
 OsBr.A8 2 O s and RbBr.As z O z . WheD the double bromides 
 are recrystallized from water, or dilute HBr, they yield these 
 oxy-compounds, and these generally separate in the form of a 
 white crust on the bottom and sides of the beaker. Analysis 
 of such products gave : 
 
 Cs . . 
 
 3CsBr.2AsBr. 
 recrystallized 
 from Water. 
 
 . . . 32.42 
 
 3CsBr.2AsBr. 
 
 recrystallized 
 from CsBr 
 Solution. 
 
 Calculated for 
 CsBr.AsjOg. 
 
 32.36 
 
 As . . 
 
 . . . 36.52 
 
 
 36.50 
 
 Br . 
 
 . . . 1957 
 
 19.59 
 
 19.46 
 
 O 
 
 ni.491 
 
 
 11.68 
 
 LbBr.2AsBr 8 
 rom Water. 
 
 Calculated for 
 EbBr.A6 2 O 8 . 
 
 16.56 
 
 23.52 
 
 50.74 
 
 41.27 
 
 15.91 
 
 22.01 
 
 [16.79] 
 
 13.20 
 
 3RbBr.2AsBr s 
 from dilute HBr. 
 
 Rb 24.24 
 
 As 40.06 
 
 Br 24.53 
 
 O 11-17 
 
 It is to be noticed that the product obtained by recrystal- 
 lizing 3RbBr.2AsBr 8 from water is impure, while the caesium 
 compound made in the same way corresponds to the formula. 
 This is an illustration of the greater tendency of the caesium 
 halide to form double salts than in the case of the rubidium 
 halide. Both these compounds are white, but the rubidium 
 compound turns somewhat yellow on drying. Under the 
 microscope, six-sided plates were seen in the case of the 
 caesium compound when the solution was slowly cooled. 
 There were also observed hexagonal crystals with a short 
 columnar rhombohedral habit. They were uniaxial with 
 weak, negative double refraction. The rubidium compound 
 was also obtained in hexagonal crystals showing rhombohedral 
 symmetry and weak negative double refraction. 
 
 CsI.AstOs, RbI.As 2 O s and KI.As 2 O s . The formation of 
 these compounds was observed when dilute hydriodic acid 
 solutions of the alkaline iodides were mixed with dilute 
 acid solutions of AsI 8 . If the solutions are mixed while hot, 
 these double salts separated on cooling in the form of crystal- 
 
 OF THE 
 
 UNIVERSITY 
 
 OF ,. 
 
 41 ICO 
 
WITH CAESIUM AND RUBIDIUM. 
 
 307 
 
 line yellow crusts on the bottom and sides of the beaker. 
 These crystals are generally somewhat larger than those of 
 the compounds of As 2 O 3 with the chlorides and bromides. 
 Under the microscope they exhibited the form of six-sided 
 plates. These show a strong negative double refraction. 
 
 The potassium compound also appeared in the form of six- 
 sided plates ; these remained dark when rotated between 
 crossed nicols. They were too small to afford an axial figure, 
 as the largest plates did not exceed 0.01 mm. in diameter. 
 They are probably hexagonal. Analyis gave : 
 
 Wnimrt Calculated for 
 
 Found. CsI.Aa,0,. 
 
 Cs 29.31 29.04 
 
 As 32.01 32.75 
 
 I 28.94 27.73 
 
 [9.74] 10.48 
 
 Calculated for 
 RbI.As 2 O 8 . 
 
 Kb 20.35 20.83 
 
 As 36.78 36.54 
 
 I 31.94 30.93 
 
 [10.93] 11.70 
 
 Calculated for 
 KI.AsjjOa. 
 
 K 10.75 10.74 
 
 As 42.85 41.20 
 
 I 34.13 34.88 
 
 [12.27] 13.18 
 
 Crystallography. 
 
 The crystallization of the caesium and rubidium arsenious 
 halides is hexagonal. They were all measured and found 
 to be isomorphous. In general the habit was holohedral, 
 although in the case of caesium arsenious bromide it is rhom- 
 bohedral. All these salts show a pronounced basal cleavage, 
 and plates parallel to this, examined with the stauroscope, are 
 uniaxial: the double chlorides and bromides show a weak 
 negative double refraction, while the double iodides are posi- 
 tive. The forms observed are as follows: 
 
308 
 
 DOUBLE HALIDES OF ARSENIC 
 
 c 0001 
 a 1150 i-2 
 m 10TO / 
 
 r 10T1 1 
 
 z 01T1 -1 
 p 2021 2 
 
 The steep pyramid p was found only on the iodides. 
 The following table gives the lengths of the vertical axes 
 
 3CsC1.2AsCl 8 
 
 3KbC1.2AsCl 8 
 
 3CsBr.2AsBr 8 
 
 3RbBr.2AsBr 8 
 
 3CsI.2AsIs 8 
 
 3BbI.2AsI, . 
 
 1.209 
 1.210 
 1.219 
 1.220 
 1.244 
 1.243 
 
 1. 
 
 A comparison of the above axial ratios shows the interest- 
 ing fact that the substitution of rubidium for caesium produces 
 no appreciable effect hi the lengths of the axes, and that in 
 this series the vertical axes lengthen as the atomic weight of 
 the halogens increases. The crystals were sufficiently stable 
 to yield good measurements, although on long exposure they 
 usually lose their lustre. In the lists of 
 measurements the angles chosen as fun- 
 damental are noted by an asterisk. 
 
 3CsCl.2AsCls. This salt was made 
 hi crystals about 1-2 mm. in diameter. 
 The forms observed are w, a, r, 2, and 
 c. A careful search was made, for indi- 
 cations of a rhombohedral development 
 of the faces r and 2, but none was 
 found. Apparently they are always holohedral in their 
 development, Fig. 1. 
 
 m 
 
 MAC, 10TOA0001 
 c A r, 0001 A 10T1 
 r AW, 11T1 A 1010 
 m A z, 10TO A 01T1 
 r A , 10T1 A 01T1 
 
 Measured. 
 90 
 
 *54 24' 
 35 39' 
 66 3' 
 47 53' 
 
 Calculated. 
 90 
 
 35 36' 
 
 66 V 
 
 47 58' 
 
 There is a rather poor prismatic cleavage, and plates parallel 
 to this show parallel extinction. 
 
WITH CAESIUM AND RUBIDIUM. 
 
 3RbCl.%AsClz. This salt was made 
 in crystals up to about 5 mm. in diameter. 
 The forms observed are <?, m, r, and 2. 
 The faces r and 2 were seldom present, 
 but when these did occur it could not be 
 seen whether they exhibited rhombohe- 
 dral symmetry or not. Penetration twins 
 are common, the twinning plane being 
 the rhombohedron 0111, Fig. 2. 
 
 Measured. 
 
 c A c (twin), 0001 A 0001 *71 3' 
 c AT-, 0001 A 1011 54 21' 
 m A r, 10TO A 1011 35 39' 
 
 Calculated. 
 
 64 
 
 35 
 
 On examining this salt in convergent polarized light a uni- 
 axial cross is seen whose arms are not black but a deep and 
 brilliant blue, the character being negative. When examined 
 in monochromatic red light, the crystals are nearly isotropic, 
 the double refraction being extremely weak and probably 
 negative. In blue light, however, a distinct cross is seen 
 accompanied by axial rings. This difference between red and 
 blue explains the colored cross seen in white light. Sections 
 parallel to the axis c show the deep peculiar blue charac- 
 teristic of uniaxial bodies with the above-mentioned optical 
 properties. 
 
 3CsBr.%AsBrz. This salt was made in crystals up to 
 mm. in diameter. The forms observed are <?, w, r, and 2. 
 
310 
 
 DOUBLE HALIDES OF ARSENIC 
 
 This is the only salt of the series that has a rhombohedral 
 habit, and as the angle of the rhombohedron is nearly 90 the 
 crystals look like cubes. Fig. 3 shows an ideal combination 
 of r with m, 2, and c. This form was not observed, as the 
 crystals are invariably twins. An ideal representation of the 
 twinning is given in Fig. 4. The r faces were so curved 
 and striated that no exact measurements could be made from 
 them. 
 
 Measured. Calculated. 
 
 m A , 01TO A 01T1 *35 23' 
 
 AZ, 10T1 A T101 ... 89 50' 
 
 z. This salt was made in small crystals, up 
 to 2 mm. in diameter. The forms observed were <?, w, and r. 
 The crystals were made in two habits. When prepared with 
 an excess of RbBr, it separates in prismatic crystals which 
 resemble the form of caesium arsenious chloride, Fig. 1. In 
 one experiment, by using an excess of AsBr 3 , contact twins 
 were obtained. Here the twinning plane is the unit rhombo- 
 hedron, as in Fig. 2, but some of the faces are lengthened par- 
 allel to the edge between the two basal planes (Fig. 5). 
 
 c AC (twin), 0001 A 0001 
 c AT, 0001 A 10T1 
 r A m, 1011 A 10TO 
 r A, lOllAOlTl 
 m A * 10TO A 01T1 
 
 Measured. 
 
 70 C 
 54 C 
 *35 C 
 48 C 
 65 C 
 
 Calculated. 
 
 70 44' 
 
WITH CAESIUM AND RUBIDIUM. 311 
 
 3CsI.%AsI z . This salt was made in beautiful crystals up to 
 3 mm. in length. The forms observed are p and <?. The habit 
 is that of a steep doubly terminated hexagonal pyramid, with 
 the basal planes small or wanting. Often the middle edges 
 are rounded by oscillatory combinations of the pyramids giv- 
 ing rise to horizontal striations. There was no indication of a 
 prism or of a rhombohedral development of the pyramidal 
 faces. 
 
 Measured. Calculated. 
 
 c*p, 0001 A 2021 *70 49' 
 
 p AJP, 2021 A 2021 38 19' 38 22' 
 
 p Aj p, 2021 A 0221 56 24J' 56 21' 34" 
 
 This salt is without optical anomalies, and no pleochroism 
 was observed. 
 
 3RbI.AsI 3 . This salt was prepared in very small crystals, 
 not over 1 mm. in length. The forms observed were c, w, and 
 p. The habit is similar to Fig. 6, but usually the middle edges 
 are replaced by the faces of the prism m or are rounded by 
 horizontal striations. 
 
 Measured. Calculated. 
 
 p A p, 2021 A 0221 *56 21' 
 
 c A p, 0001 A 2021 70 47' 70 
 
 p A m, 2021 A 10TO 19 12' 20 56' 
 
 Optically this salt shows anomalies. Basal cleavage plates 
 between crossed nicols are not dark, but light and remain so 
 during revolution. In convergent light the locus of an optic 
 axis is seen in the centre of the field coinciding with the ver- 
 tical axis <?. The bisectrix lying nearest this axis is that of 
 least elasticity, C. The salt is not, however, properly biaxial ; 
 the plane of the optic axes is sometimes parallel, sometimes 
 perpendicular to the edge of the prism. Moreover this direc- 
 tion often changes from place to place in the same plate, and 
 at times no bar is seen, but a black dot in the centre of the 
 field surrounded by rings. Such behavior can be explained by 
 supposing the crystal to be in a condition of internal strain. 
 Sections parallel to the prism sometimes remain light between 
 crossed nicols, sometimes extinguish at varying angles and 
 
312 DOUBLE HALIDES OF ARSENIC. 
 
 show a slight pleochroism, the absorption being e > o; the 
 color a deep reddish orange. 
 
 In conclusion the author wishes to express his indebtedness 
 to Prof. H. L. Wells for valuable advice in connection with 
 the present investigation, and to Prof. S. L. Penfield, under 
 whose direction the crystallography of these salts was investi- 
 gated. The author is also indebted to Mr. L. V. Pirsson for 
 aid in the optical description of these salts. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 March, 1893. 
 
ON SOME DOUBLE SALTS OF LEAD 
 TETRACHLORIDE.* 
 
 BY H. L. WELLS. 
 
 THE existence of lead tetrachloride has long been surmised 
 from the fact that the corresponding oxide, when dissolved in 
 cold hydrochloric acid, gives a yellow solution in which sul- 
 phuric acid does not give an immediate precipitate. Lead 
 tetrachloride itself, however, has never been isolated, nor has 
 any double salt which it forms been satisfactorily described. 
 
 Sobrero and Selmi f found that when chlorine is passed into 
 a solution containing sodium chloride and lead chloride, the 
 liquid becomes yellow. They found it impossible to isolate 
 the compound either by evaporation or cooling, so that they 
 determined the lead, sodium, and chlorine in such a solution, 
 and found it to contain these constituents in the ratio corre- 
 sponding to PbCl 4 + 9NaCl. Sobrero and Selmi say that per- 
 haps this is the formula of the compound, but they put a 
 question-mark after it. Their analysis indicates the existence 
 of PbCl 4 in combination with NaCl, but if the solution had 
 contained a compound of that composition, which was stable 
 with water, it probably could have been isolated by evapora- 
 tion. The fact is that the double salts of lead tetrachloride are 
 not stable with water, as will be shown in the present article. 
 Therefore, since a large excess of sodium chloride must have 
 been present in the solution of Sobrero and Selmi, their analy- 
 sis could not have determined the composition of the double 
 salt that it contained. 
 
 Nickles* saturated a strong solution of calcium chloride with 
 lead chloride and chlorine and analyzed the solution. He 
 
 * Amer. Jour. ScL, xlvi, September, 1893. 
 t Ann. Chim. Phys., Ill, xxix, 161. 
 
 * Ibid., IV, x, 323. 
 
314 ON SOME DOUBLE SALTS 
 
 found it to contain lead, calcium, and chlorine in the propor- 
 tions represented by PbCl 4 + 16CaCl 2 . In conclusion NicklSs 
 does not claim that any such double salt exists, but merely 
 claims to have indicated the existence of PbCl 4 . 
 
 In view of the fact that the formulae PbCl 4 + 9NaCl and 
 PbCl 4 -f 16CaCl 2 merely represented the composition of solu- 
 tions, it is remarkable that they are given in some handbooks 
 of chemistry as real chemical compounds. It may be men- 
 tioned that Carnegie* has used the formula PbCl 4 .9NaCl in 
 support of a theory on double halides. 
 
 O. Seidel f mentions unsuccessful attempts to isolate PbCl 4 
 and its double salts with the chlorides of other metals. 
 
 Fisher J dissolved lead peroxide in hydrochloric acid, and 
 found that all the lead in the solution was precipitated again 
 as peroxide by the addition of sodium acetate. He was evi- 
 dently not aware of the fact that Rivot, Beudant, and Daguin 
 had shown, long before, that lead is completely precipitated as 
 peroxide by the addition of sodium acetate and chlorine to its 
 solutions. Fisher found that two atoms of chlorine were used 
 (as would be expected) in precipitating one atom of lead as 
 peroxide. His conclusion that his experiments showed the 
 existence of lead tetrachloride has, apparently, little foundation. 
 
 More recently, Ditte || has made some experiments on the 
 solubility of lead chloride in solutions containing hydrochloric 
 acid and chlorine. He apparently does not believe in the 
 existence of lead tetrachloride, for he does not mention the 
 compound, while he explains the precipitation of lead peroxide, 
 when such solutions are diluted, by saying that lead chloride 
 is partly dissociated by the act of solution, that the solution 
 then contains oxide of lead, and that this is peroxidized by the 
 oxides of chlorine formed when chlorine is passed into the 
 solution. 
 
 Nikolukine has succeeded in isolating double salts of lead 
 tetrachloride with ammonium and potassium chlorides. He 
 
 * Am. Chem. Jour., xv, 10, 1893. t Jour. pr. Ch., II, xx, 205, 1879. 
 
 t Jour. Chem. Soc., xxxv, 282, 1879. Ann. Mines, V, iv, 239, 1853. 
 || Ann. Chim. Phys., V, xxii, 566, 1881. 
 
OF LEAD TETRACHLORIDE. 315 
 
 showed that these compounds contain lead and extra chlorine 
 in the proportion required for PbCl 4 , but there is no evidence 
 in the abstracts of mYarticle * that he determined the composi- 
 tion of the double salts. His original article in Russian is not 
 accessible to me. Nikolukine prepared the compounds by 
 dissolving lead dioxide in concentrated hydrochloric acid in 
 sealed tubes, and adding the alkaline chlorides to the solutions 
 thus produced. He describes the double salts as having a 
 lemon-yellow color, and states that they are pretty stable, the 
 ammonium chloride compound being decomposed at 120. 
 
 The present investigation has been undertaken with the view 
 of determining the composition of the salts which Nikolukine 
 discovered, and especially in order to investigate the corre- 
 sponding rubidium and caesium compounds, which, from anal- 
 ogy, were expected to be more insoluble and stable than the 
 potassium salt. As a result, it has been found possible to pre- 
 pare the whole series in a state of purity, and the expectations 
 in regard to the easy preparation of the rubidium and caesium 
 salts have been fully realized. The following salts are to be 
 described : 
 
 (NH 4 ) a PbCl, 
 
 K 2 PbCl 6 
 
 Eb 2 PbCl 6 
 
 Cs 2 PbCl 6 
 
 These salts are all yellow, and they all crystallize in the 
 isometric system with an octahedral habit. 
 
 These salts show a new relation between lead and other 
 metals of Mendele'eff's group IV, with which this type is very 
 common, especially among the double fluorides. It is to be 
 noticed also that this type is almost invariable among the 
 double salts which tetrahalides form, for platinum, iridium, 
 osmium, and palladium give analogous, isomorphous com- 
 pounds, while, as has been recently shown by Dr. H. L. 
 Wheeler of this laboratory, tellurium gives an extensive series 
 of octahedral salts of this type. The octahedral form of the 
 
 * Berichte, xviii, 370. 1885 ; Jour. Chem. Soc., i, 123. 
 
316 ON SOME DOUBLE SALTS 
 
 anhydrous salts of this type is very characteristic, and it seems 
 to be universal, except among the fluorides. 
 
 All of the lead salts to be described are decomposed by water 
 with the formation of lead peroxide. Chlorine is usually set 
 free at the same time. It may be assumed that two successive 
 reactions take place, which may be represented by the following 
 equations : 
 
 (1) PbCl 4 + 2H 2 = Pb0 2 + 4HC1 
 
 (3) Pb0 2 + 4HC1 = PbCl 2 + C1 2 + 2H 2 
 
 The extent to which the second reaction takes place depends 
 upon the dilution and the temperature. If the amount of 
 water present is not too great, a state of equilibrium is reached 
 when a sufficient amount of alkaline chloride, hydrochloric 
 acid, and chlorine have gone into solution, and the decomposi- 
 tion stops. The caesium salt is more slowly decomposed by 
 water than others. All the salts are decomposed by boiling 
 with an excess of hydrochloric acid, but the decomposition of 
 the caesium compound is remarkably slow, especially in solu- 
 tions containing much csesium chloride. 
 
 When free chlorine is present the csesium salt is almost 
 completely insoluble in strong solutions of csesium chloride and 
 in hydrochloric acid. Although the rubidium salt is consider- 
 ably more soluble, the difference is not great enough so that a 
 quantitative separation can be made. It will be shown in the 
 following article that caesium can be approximately separated 
 from potassium, sodium, and lithium by this means, and that 
 when rubidium is also present the csesium can be approxi- 
 mately determined indirectly. 
 
 The salts to be described can be washed with hydrochloric 
 acid containing chlorine. They are perfectly stable on expo- 
 sure to the air. When heated in capillary tubes, the ammonium 
 salt begins to whiten at about 225, the potassium salt at about 
 190, and the csesium and rubidium salts at about 280. This 
 temperature for the decomposition of the ammonium salt is 
 about 100 higher than that given by Nikolukine. It is prob- 
 able that difference is due to a typographical error. 
 
OF LEAD TETRACHLORIDE. 317 
 
 Attempts were made to prepare corresponding sodium and 
 calcium salts, without success. 
 
 In analyzing the salts, lead was separated and weighed as 
 sulphate, and, in the filtrate from this, the alkali metal was 
 determined as sulphate. To determine chlorine, a separate 
 portion was decomposed by a solution of sodium arsenite and 
 chlorine was determined in this as usual. 
 
 Ammonium-Plumbic Chloride, (NH^) z PbCl & . In prepar- 
 ing this salt, Nikolukine's method of using sealed tubes was 
 found to be unnecessary. A solution of lead tetrachloride was 
 made by adding slightly diluted hydrochloric acid to an excess 
 of lead dioxide at 0. This solution was quickly filtered 
 through asbestos, and a saturated, cold solution of ammonium 
 chloride in dilute hydrochloric acid was added until an abun- 
 dant, yellow, crystalline precipitate was produced. The salt 
 was pressed on paper, and then air-dried. 
 
 v~ *A Calculated for 
 
 (NH 4 ) 2 PbCl. 
 
 Ammonium 7.90 
 
 Lead 44.61 45.39 
 
 Chlorine 46.53 45.71 
 
 Potassium-Plumbic Chloride, K 2 PbCl 6 . Chlorine was 
 passed into a solution saturated with potassium chloride, lead 
 chloride, and hydrochloric acid at 0, without producing the 
 double salt. Nikolukine has stated that the salt is soluble in 
 an excess of potassium chloride, and, acting upon this sugges- 
 tion, another solution was made, like the former except that 
 no potassium chloride was used. On mixing about equal vol- 
 umes of the two solutions and letting the mixture stand at 
 for several hours, a well-crystallized crop of the yellow double- 
 salt was obtained. The air-dry salt was analyzed. 
 
 Found. Cal dor 
 
 Potassium 15.30 
 
 Lead 41.91 
 
 Chlorine 42.49 
 
 9070 
 Loss on heating .... 15.07 C1 2 14.25 
 
318 ON SOME DOUBLE SALTS 
 
 The above method of preparation gives a small yield, and it 
 would probably be better to use a method analogous to that by 
 which the ammonium salt was prepared. 
 
 Rubidium-Plumbic Chloride, Rb 2 PbCl 6 . When 65 g. of 
 rubidium chloride were dissolved in 250 c. c. of water with 
 4 g. of lead chloride, no precipitate was produced by saturating 
 the solution with chlorine, but, on adding an equal volume of 
 concentrated hydrochloric acid to this solution, an abundant, 
 yellow, crystalline precipitate was produced. This was col- 
 lected on a filter, washed with hydrochloric acid containing 
 chlorine, and air-dried. 
 
 Wrt , j Calculated for 
 
 Rubidium ...... 28.62 28.93 
 
 Lead ....... 34.98 35.03 
 
 Chlorine ...... 35.85 36.03 
 
 9045 100.00 
 
 Loss on heating .... 12.41 C1 2 12.01 
 
 A solution 35 c. c. in volume, made of equal volumes of 
 concentrated hydrochloric acid and water, and containing 
 .0619 g. of rubidium and double the theoretical quantity of 
 lead chloride, was saturated with chlorine. A precipitate of 
 the double salt was produced, which, after standing several 
 hours, was collected upon a Gooch filter. The rubidium in 
 this precipitate was determined and found to amount to .0318 g. 
 One cubic centimeter of the solution dissolved, therefore, 
 .003 g. of the lead salt, equivalent to .00086 g. of rubidium. 
 The experiment was made at about 20. 
 
 Ccesium- Plumbic Chloride, Cs PbCl*. This salt is very 
 readily prepared by passing chlorine into solutions containing 
 lead chloride and a large excess of caesium chloride. When 
 hydrochloric acid is present, the excess of csesium chloride is 
 unnecessary, but in that case the precipitate is very finely 
 divided. The precipitate begins to form in solutions that are 
 nearly at a boiling temperature. A crop obtained without the 
 use of hydrochloric acid was analyzed. It was washed with 
 hydrochloric acid containing chlorine and air-dried. 
 
OF LEAD TETRACHLORIDE. 
 
 319 
 
 Found. 
 
 Caesium 38.51 
 
 Lead 30.05 
 
 Chlorine 30.99 
 
 99.55 
 
 Loss on heating . . . 10.96 
 
 ci, 
 
 Calculated for 
 Cs 2 PbCl 6 . 
 
 38.78 
 30.17 
 31.05 
 100.00 
 10.35 
 
 The salt usually has a lemon-yellow color, but, when very 
 strong hydrochloric acid is used and a large excess of lead 
 chloride is present, the precipitate has a dark brown color. 
 Such a crop gave the following analysis : 
 
 Found. 
 
 Caesium 38.19 
 
 Lead 29.64 
 
 Chlorine 31.35 
 
 99.18 
 11.09 
 
 Calculated for 
 Cs 2 PbCl. 
 
 38.78 
 30.17 
 31.05 
 100.00 
 C1 2 10.35 
 
 Loss on heating . . . 
 
 This is evidently the same compound 
 as the lemon-yellow salt. The cause of 
 the brown color is not known. The 
 presence of lead dioxide in it does not 
 seem probable on account of the strong 
 acid that was used, and, moreover, ex- 
 periment showed that this oxide was 
 instantly dissolved by the mother-liquor. 
 It was suspected that this was a dimorphous form of the 
 compound, but Mr. Louis V. Pirsson, who has kindly made 
 a microscopic examination of both products, has found that 
 both are isometric and octahedral in habit. He noticed that 
 while the yellow salt forms perfect octahedrons, the brown 
 compound occurs in octahedral groups composed of combina- 
 tions of the cube and octahedron. The accompanying figure, 
 by Mr. Pirsson, shows the prevailing habit of these crystals. 
 The groups are very small, usually not over 0.015 mm. 
 in diameter. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 March, 1893. 
 
ON THE DOUBLE HALIDES OF ANTIMONY 
 WITH RUBIDIUM.* 
 
 BY H. L. WHEELER. 
 
 THE investigations of the double halides of antimony and 
 rubidium have hitherto been confined to the chlorides, and the 
 following salts have been described : 
 
 1 : 1 Eubidium Antimony Chloride, EbCl.SbCl 8 . 
 
 5:3 5RbC1.3SbCl 8 
 
 23 : 10 " " " 23RbC1.10SbCl 8 
 
 6:1 " " " 6RbCl.SbCl 8 
 
 The first three of these compounds were described by Remsen 
 and Saunders.f These investigators, after a careful study of 
 the subject, came to the conclusion that the salt 6RbCl.SbCl 8 
 described by GodeffroyJ does not exist. 
 
 It has been shown by the author of the present article that 
 the 3 : 2 type of double salts is probably the only one formed 
 by the combination of the arsenic halides with those of csesium 
 and rubidium. Moreover, since this type was observed by 
 Sch8effer|| in the case of the double halides of antimony with 
 sodium, potassium, and ammonium, and, since Remsen and 
 Saunders obtained the salt 3CsC1.2SbCl 3 it seemed probable 
 that this type of double halides would exist with rubidium 
 and antimony. A thorough re-examination of the chlorides 
 has therefore been undertaken, and the investigation has been 
 extended to the bromides and iodides. As a result of this 
 investigation the following compounds have been obtained : 
 
 KbC1.2SbCl 8 .H 2 
 
 KbCl.SbCl 8 . 
 
 3RbC1.2SbCl 8 . 3KbBr.2SbBr 8 3RbI.2SbI 8 
 
 23RbC1.10SbCl 8 (?) 23RbBr.lOSbBr 8 (?) 
 
 * Amer. Jour. Sci., xlvi, Oct., 1893. t Amer. Chem. Jour., xiv, 155. 
 $ Berichte, viii, 9. Amer. Jour. Sci., xlvi, 88. 
 
 II Pogg. Ann., cix, 611. 
 
DOUBLE HALIDES OF ANTIMONY. 321 
 
 The first chloride, RbC1.2SbCl 8 .H 2 O, is a new type of 
 antimony rubidium halides, which Remsen and Saunders did 
 not obtain. The second, 1:1, confirms the results of these 
 investigators, while the series of 3 : 2 salts, which includes a 
 chloride, bromide, and iodide, corresponds to the type expected 
 from analogy. The difference between the percentage compo- 
 sition required for the 3 : 2 chloride and that required for the 
 5 : 3 formula of Remsen and Saunders is small, and it is to be 
 noticed that these authors do not consider their formula as 
 definitely established. They say, " The analytical results ob- 
 tained from different samples varied considerably and it does 
 not appear possible to obtain the salt in pure condition." It 
 will be noticed that most of the analyses of the 3 : 2 chloride, 
 made in the present investigation, show a composition inter- 
 mediate between what is required for the formulas of the 3 : 2 
 and the 5 : 3 salts, but the bromide and the iodide were readily 
 obtained in pure condition and gave analytical results closely 
 corresponding to the 3:2 formula. Moreover the chloride, 
 bromide, and iodide just mentioned are all hexagonal and may 
 be referred to axes which correspond closely to those of the 
 3 : 2 arsenic compounds. The chloride and bromide with a 
 complex composition (23 : 10 ?) confirm the results of Remsen 
 and Saunders on the chloride. The formula suggested by 
 them has been retained, subject to uncertainty. It will be 
 seen beyond that, as Remsen and Saunders have noticed, 
 the ratio 16 : 7 corresponds very closely to the analyses, and 
 it may be added that the ratios 9 : 4 and 7 : 3 differ so little 
 from the other two that it would be very difficult to dis- 
 tinguish between any of these ratios by analysis. 
 
 For the preparation of the double halides the constituents 
 were mixed in the presence of the corresponding dilute acids. 
 In the case of the chlorides a 10 per cent acid was used. 
 The mixtures were then evaporated until crystals separated 
 on cooling. Further details will be given with the descrip- 
 tions of the salts. In the case of each salt several crops 
 were prepared and analyzed, and an attempt has been made 
 to determine approximately the limits of the conditions under 
 
 21 
 
322 ON THE DOUBLE HALIDES OF 
 
 which these double halides are formed. It may be added that 
 the analytical results are not selected, for with the exception 
 of two antimony determinations, where an error had been 
 detected, every determination that was made has been given. 
 
 Method of Analysis. 
 
 The salts were removed from the mother-liquor, and, after 
 pressing on smooth filter paper, were dried in the air for a 
 short time. Portions of a little less than half a gram were 
 taken for analysis. In order to determine the halogens, silver 
 nitrate was added to a solution of the substance in water 
 containing a little tartaric and nitric acids, the mixture was 
 then warmed on the water bath for a couple of hours, and 
 finally, after standing twelve hours, the silver halide was col- 
 lected, ignited, and weighed in a Gooch crucible in the usual 
 manner. The determination of the antimony and rubidium 
 was effected in a separate sample. In order to do this, the 
 salts were dissolved in a little dilute hydrochloric acid and the 
 solutions were diluted with boiling water. Hydrogen sulphide 
 was then used to precipitate the antimony, and, when the solu- 
 tions had cooled, the resulting sulphide was filtered on asbes- 
 tos in a Gooch crucible, washed with water and alcohol and 
 then heated to 230 in an oven filled with carbonic acid. On 
 cooling, the sulphide was weighed as Sb 2 S 8 . The rubidium 
 was determined by evaporating the filtrate from the antimony 
 sulphide to dryness with an excess of sulphuric acid, the 
 residue was then converted into normal sulphate by ignition 
 in a stream of air containing ammonia. The atomic weights 
 used in the calculation of results were the following : 
 
 Cl, 35.5 ; Br, 80 ; I, 127 ; Sb, 120 ; Rb, 85.5 
 
 The Double Chlorides. 
 
 The crystals of the double chlorides are colorless, with the 
 exception of the salt 3RbC1.2SbCl 8 ; this salt has a pale yellow 
 color exactly similar to the salts 3RbC1.2AsCl 8 and 3CsCl. 
 2AsCl 8 . The stability of the double chlorides, on exposure, 
 
ANTIMONY WITH RUBIDIUM. 323 
 
 appears to vary inversely with the quantity of antimony chlo- 
 ride which they contain. 
 
 1:2 Rubidium Antimony Chloride, RbCl.2SbCl t .]I t O.- 
 This new salt was obtained from hydrochloric acid solution 
 when the constituents were mixed in the proportion of ten, 
 eight, or six molecules of SbCl 3 to one of RbCl. On concen- 
 trating these mixtures supersaturated solutions were obtained 
 which sometimes remained for days without giving crystals, 
 but on shaking, or stirring with a glass rod, the crystallization 
 was induced. The crystals separate in the form of elongated, 
 colorless, monoclinic tables. Analysis of different crops gave : 
 
 From Solutions of 
 10SbCl s to IRbCl. 
 
 From Solution 
 of 8SbCl 3 to 
 IRbCl. 
 
 From 
 Solution 
 of 6SbCl 3 
 to IBbCl. 
 
 Calculated for 
 RbC1.2SbCl s .H,O. 
 
 Kb 
 
 14.61 
 
 14.71 
 
 14.74 
 
 14.64 
 
 15.07 
 
 14.44 
 
 Sb 
 
 40.75 
 
 40.97 
 
 41.09 
 
 41.07 
 
 40.97 
 
 40.54 
 
 Cl 
 
 41.83 
 
 41.53 
 
 41.11 
 
 . . . 
 
 . . . 
 
 41.98 
 
 H 2 
 
 3.20 
 
 3.10 
 
 3.18 
 
 3.08 
 
 
 
 3.04 
 
 The crystals of this salt have a brilliant lustre when first re- 
 moved from the mother-liquor, but on exposure they soon 
 lose their lustre, becoming opaque and decomposing. In the 
 preparation of this salt for analysis the crystals were crushed 
 and thoroughly pressed on filter paper, and when it was certain 
 that the powder did not contain any mechanically mixed water, 
 it was placed in a weighing-tube. This salt is readily dis- 
 tinguished from the other colorless double halides of rubidium 
 and antimony by the fact that it melts at 77. 
 
 1 : 1 Rubidium Antimony Chloride, RbC2.SbCl s . This salt 
 was first described by Remsen and Saunders ; * they say that 
 " if the excess of antimony chloride ... be very great, a 
 colorless salt crystallizing in elongated, apparently orthorhom- 
 bic, crystals is obtained." I have found that by mixing the 
 constituents in hydrochloric acid solutions, in the proportion 
 of four or three molecules of SbCl 8 to one of RbCl, crystals of 
 similar appearance were obtained. The solutions require a 
 
 * Loc. cit. 
 
324 ON THE DOUBLE HALIDES OF 
 
 considerable degree of concentration, and the mother-liquor is 
 more or less syrupy, hence the rubidium determinations came 
 low and the antimony high. Analysis gave : 
 
 From Solution of From Solution of Calculated for 
 
 4SbCl 3 to IRbCl. 3SbCl 3 to IRbCl. KbCl.SbCl 3 . 
 
 Eb 23.67 23.96 24.61 
 
 Sb 35.38 34.99 34.53 
 
 Cl 40.70 40.73 40.86 
 
 A solution of antimony and rubidium chloride in the pro- 
 portion of 2^ molecules of the former to one of the latter 
 gave a mixture of this salt and the yellow one described 
 below. As has been observed by Remsen and Saunders, 
 crystals of this salt rapidly lose their lustre on exposure. 
 They give no definite melting-point below the temperature of 
 boiling sulphuric acid. 
 
 3\% Rubidium Antimony Chloride, 3Rt>ClSbCl 9 . Tbis 
 is the salt to which Remsen and Saunders assign the formula 
 5RbC1.3SbCl 8 . They obtained this compound on adding " a 
 considerable excess " of antimony chloride to a solution of the 
 salt 23RbC1.10SbCl 3 . They describe the crystals as some- 
 times resembling a rhombohedron in general shape and having 
 a pale yellow color, and they remark that " this is noteworthy, 
 because the more complex salt (23RbC1.10SbCl 8 ) and the 
 simpler one (RbCl.SbCl 3 ) are both colorless. It is to be 
 remembered, however, that the salt Cs 3 Sb 2 Cl 9 (3CsC1.2SbCl 8 ) 
 is also yellow." It may be added that both 3CsCl. 2AsCl 3 and 
 3RbC1.2AsCl 8 are pale yellow. Remsen and Saunders also 
 remark: "As the formula of this rubidium salt is not very 
 simple, and as the substance could not be recrystallized, on 
 account of the strong tendency towards the formation of the 
 very complex salt, the formula suggested below can hardly be 
 considered as definitely established." 
 
 I have found that when solutions of antimony chloride and 
 rubidium chloride are mixed in the proportion of one and one- 
 fifth molecules of the former to one molecule of the latter a 
 pale yellow salt is obtained crystallizing in rhombohedra. 
 
ANTIMONY WITH RUBIDIUM. 
 
 325 
 
 In one case, on obtaining a crop of crystals from a solution 
 of 2SbCl 8 to IRbCl in strong HC1, the yellow rhombohedra 
 were seen to be mixed with the colorless hexagonal plates, 
 presumably of the salt 23RbC1.10SbCl 8 . It was also found 
 that a wide difference exists in the solubility of these two 
 salts in warm solutions ; the yellow crystals dissolved with 
 difficulty, while on the other hand the salt 23 : 10 went into 
 solution with only a slight elevation of temperature. If the 
 crystals of the yellow salt are warmed in the mother-liquor 
 they become opaque throughout without losing their pale yel- 
 low color. It seems probable that impurities are dissolved out 
 by this operation and that no decomposition takes place, for 
 the decomposition products and the other double chlorides are 
 colorless. An analysis of a crop obtained in this manner cor- 
 responded very closely to the formula 3RbC1.2SbCl 8 . Analy- 
 sis gave : 
 
 From 
 Solutions of 
 2SbCl s to 
 IRbCL 
 
 From 
 Solutions of 
 l$SbCl 3 to 
 IRbCl. 
 
 From 
 Solutions of 
 2SbCl 3 to 
 IRbCl 
 heated. 
 
 Calculated 
 for 
 3RbCl. 
 2SbCl8. 
 
 Calculated 
 for 
 5RbCl. 
 3SbCl 8 . 
 
 Kb 
 
 32.57 
 
 32.19 
 
 33.34 
 
 31.86 
 
 31.30 
 
 31.44 
 
 33.28 
 
 Sb 
 
 28.68 
 
 28.67 
 
 28.55 
 
 28.46 
 
 29.44 
 
 29.41 
 
 28.03 
 
 Cl 
 
 38.38 
 
 38.42 
 
 38.32 
 
 ... 
 
 38.98 
 
 39.15 
 
 38.69 
 
 23:10? Rubidium Antimony Chloride, 
 - For the preparation of this compound, a sample of rubidium 
 chloride was used which had been specially purified for the 
 purpose by the method recently described by Prof. H. L. 
 Wells* of this laboratory. The purification of this sample 
 was repeated after the product failed to give spectroscopic 
 reactions for potassium and caesium. 
 
 If solutions of antimony and rubidium chlorides are mixed 
 in the proportion of one molecule of SbCl to one, four, or 
 six molecules of RbCl, the crystals obtained are the " color- 
 less six-sided plates, tables, or thicker crystals," to which 
 Remsen and Saunders have assigned the formula 23RbCl. 
 10SbCl 8 . The average results of the analyses of the different 
 
 * Amer. Jour. Sci., HI, xlvi, 188. 
 
326 ON THE DOUBLE HALIDES OF 
 
 crops of the double chloride gave figures closely agreeing with 
 those of the above authors, but the ratio of rubidium to anti- 
 mony came somewhat lower than theirs. Analysis gave : 
 
 From 
 Solution 
 6RbCl to 
 
 From 
 Solution 
 4RbCl to 
 
 From 
 Solution 
 IRbCl to 
 
 Sample 
 recrystal- 
 lized from 
 
 Average. 
 
 Ratio. 
 
 18bCl 3 . 
 
 18bCl 3 . 
 
 18bCl 3 . 
 
 10% HCL 
 
 
 
 Eb 
 
 38.98 
 
 38.55 
 
 38.83 
 
 38.62 
 
 38.60 
 
 38.716 
 
 2.28 
 
 Sb 
 
 23.76 
 
 
 23.98 
 
 23.52 
 
 23.81 
 
 23.767 
 
 1.00 
 
 m 
 
 37.16 
 
 36.97 
 
 
 
 36.95 
 
 37.026 
 
 5.26 
 
 culated for 
 )C1.10SbCl s . 
 
 38.96 
 
 Calculated for 
 !GRbC1.7SbCl 3 . 
 
 38.85 
 
 Calculated for 
 9RbC1.4SbCl 8 . 
 
 38.57 
 
 Calculated for 
 7RbC1.3SbCl 3 . 
 
 39.21 
 
 23.77 
 
 23.86 
 
 24.06 
 
 23.58 
 
 37.27 
 
 37.29 
 
 37.37 
 
 37.21 
 
 Kb . 
 
 Sb . 
 Cl . 
 
 It is to be noticed that this salt is formed under conditions 
 varying more widely than in the case of any of the other double 
 rubidium antimony chlorides. It can be exposed to the air for 
 several days without losing its lustre; on long exposure it 
 becomes covered with a white, opaque layer, probably of anti- 
 mony oxychloride. 
 
 The Double Bromides. 
 
 The bromides were obtained in the form of brilliant yellow, 
 six-sided plates, resembling the double arsenic bromides of 
 rubidium and caesium. They are comparatively stable in the 
 air, but on long exposure the crystals lose their lustre. 
 
 3 : 2 Rubidium Antimony Bromide, 3RbBr.2SbBr y This 
 salt was obtained from dilute hydrobromic acid solutions when 
 the constituents were mixed in the proportion of two and three- 
 tenths and also four molecules of RbBr to one of SbBr 3 ; it 
 was also the only one formed when antimony bromide was 
 present in the solutions in excess. It will be seen that a much 
 larger range of conditions exists for the preparation of the salt 
 3RbBr.2SbBr 8 than in the case of the corresponding double 
 chloride. Moreover, the bromide can be recrystallized unaltered 
 from dilute hydrobromic acid. 
 
ANTIMONY WITH RUBIDIUM. 
 
 327 
 
 Analysis gave : 
 
 From Solutions 
 containing a 
 large Excess of 
 SbBr 3 . 
 
 From 
 Solution 
 23RbBr. 
 to 10SbBr 3 . 
 
 From 
 Solution 
 4RbBr to 
 !SbBr s . 
 
 Sample 
 of Latter 
 recrystal- 
 lized from 
 HBr. 
 
 Calculated, 
 for 
 SRbBr. 
 
 2SbBr 3 . 
 
 21.55 
 
 21.18 
 
 20.96 
 
 21.16 
 
 21.53 
 
 20.92 
 
 21.08 
 
 . 
 
 20.07 
 
 20.13 
 
 19.98 
 
 19.59 
 
 19.91 
 
 19.73 
 
 59.30 
 
 
 
 
 
 59.07 
 
 59.19 
 
 Rb 
 Sb 
 Br 
 
 23 : 10 (?) Rubidium Antimony Bromide, 23 RbBr. 10 SbBr & . 
 This salt was obtained when dilute hydrobromic acid solu- 
 tions of rubidium and antimony bromides were mixed in the 
 proportion of six, eight, and thirteen molecules of the former 
 to one of the latter. The crystals obtained on slowly cooling 
 these mixtures, with the exception of their strong yellow color, 
 closely resemble the corresponding complex chloride. If the 
 solutions are rapidly cooled the salt separates in the form of 
 brilliant spangles. The average of the following results gives 
 a remarkably close ratio to that required for the formula 
 23RbBr.lOSbBr 8 . 
 
 Analysis gave: 
 
 Rb 
 Sb 
 Br 
 
 Solution 
 GRbBr to 
 ISbBrg. 
 
 26.66 
 16.11 
 
 Solution 
 SRbBr to 
 18bBr s . 
 
 26.16 
 
 16.23 
 
 From Solution 
 ISRbBr to ISbBr,. 
 
 Sample 
 of Latter 
 recrystallized 
 from 
 cone. HBr. 
 
 26.39 
 16.18 
 57.41 
 
 26.92 
 16.18 
 57.27 
 
 26.60 
 16.26 
 57.23 
 
 26.71 
 16.22 
 
 Rb 
 
 Calculated for 
 23RbBr.lOSbBr 8 . 
 
 . . 26.55 
 
 Calculated for 
 16RbBr.7SbBr 8 . 
 
 26.47 
 
 Calculated for 
 9RbBr.4SbBr 8 . 
 
 26.27 
 
 Calculated for 
 7RbBr.3SbBr s . 
 
 26.74 
 
 Sb 
 
 . . 16.20 
 
 16.25 
 
 16.38 
 
 16.08 
 
 Br 
 
 57.25 
 
 57.28 
 
 57.35 
 
 57.18 
 
 Average of 
 
 analytical Ratios derived 
 
 Results. 
 
 Rb . . 26.57 .3107 or 23.03 or 16.12 or 9.21 or 6.90 
 Sb . . 16.19 .1349 " 10.00 " 7.00 " 4.00 " 3.00 
 Br . . 57.30 .7162 53.09 37.16 " 21.23 " 15.92 
 
 It is to be noticed that this salt is formed within a much 
 smaller range of conditions than in the case of the chloride, and 
 
328 ON THE DOUBLE HALIDES OF 
 
 can only be recrystallized from strong hydrobromic acid solu- 
 tions. When recrystallized from moderately strong acid a 
 mixture of the salts 23 : 10 and 3 : 2 was obtained, but from 
 dilute acid a pure crop of the 3 : 2 compound separated. 
 
 Recrystallized Recrystallized Calculated for 
 
 from strong from dilute SRbBr l) SbBr 
 
 HBr. HBr. 
 
 Kb 25.54 21.89 21.08 
 
 Sb 16.93 19.70 19.73 
 
 Br 57.77 . . . 59.19 
 
 The Double Iodide. 
 
 3:2 Rubidium Antimony Iodide, 3R.bI.2SbI z . The for- 
 mation of this salt was observed when a solution of rubidium 
 iodide in hydriodic acid was saturated hot with antimony 
 iodide ; it was also obtained from a solution of antimony iodide 
 in a large excess of rubidium iodide. The best crystals are 
 obtained when a considerable quantity of antimony iodide is 
 present ; under these conditions large deep red lozenge-shaped 
 crystals separate. Analysis gave : 
 
 Large Excess Large Excess Calculated for 
 
 of Rbl. of SbI 8 . 3KbI.'28bI 8 . 
 
 Kb 16.28 14.82 15.64 
 
 Sb 14.14 15.17 14.64 
 
 I 69.76 69.55 69.72 
 
 On exposure to the air the crystals slowly lose their lustre. 
 
 Crystallography. 
 
 The crystallization of the 3 : 2 double salts is hexagonal. 
 In general the habit is rhombohedral and they can all be re- 
 ferred to axes of nearly equal length. The double bromide 
 and iodide have a perfect basal cleavage, like the salts of the 
 arsenic series, while the chloride gave only a conchoidal frac- 
 ture. The axial ratios of the salts is shown by the following 
 table, the ratios of the corresponding arsenic salts being given 
 for comparison. 
 
ANTIMONY WITH RUBIDIUM. 
 
 329 
 
 a : c a : c arc 
 
 3RbC1.2SbCl 3 1 : 1.125 3RbC1.2AsCl 3 1 : 1.210 3CsC1.2AsCl 8 1 : 1.209 
 
 3RbBr.2SbBr 8 1 : 1.207 3RbBr.2AsBr 3 1 : 1.220 3CsBr.2AsBr 8 1 : 1.219 
 
 3RbI.2SbI 8 1 : 1.230 3RbI.2AsI 3 1 : 1.242 3CsI.2A8l 3 1 : 1.244 
 
 From this table it may be seen that the substitution of 
 arsenic by antimony produces little if any effect in the lengths 
 of the axes, and in each series 
 the vertical axes lengthen as the 
 atomic weight of the halogen 
 increases. 
 
 3Kb Cl.2SbCl 3 . This salt, un- 
 like the others of the series, shows 
 rhombohedral tetartohedrism. In 
 one crop, where the crystals meas- 
 ured 5 to 7 mm. in diameter, the 
 
 faces were developed on every crystal as in Fig. 1, the forms 
 being 
 
 a, 1120, i-2 e, 01T2, - y, 2532, 
 
 Fig. 1. 
 
 m, 10TO, / 
 
 v, T322, 
 
 On a second crop only e and a were developed. 
 
 Calculated. 
 
 e A e, 01T2 A 1T02 
 e A v, 01T2 A T322 
 e A ?/, T322 A 2532 
 y A a, 25H2 A T2TO 
 
 Measured. 
 
 56 18' 
 29 53' 
 11 29' 
 
 20 29' 
 
 29 55' 
 11 26' 
 20 29' 
 
 This salt differs from all the others of the series, since it is 
 the only one on which tetartohedrism has been observed. 
 Whether the others are really tetartohedral, but have not 
 shown it owing to the absence of highly modified forms, can- 
 not be told at present. Also the basal cleavage, which is so 
 prominent on all of the others, could not be detected, while 
 the one-half rhombohedron e was only observed on this salt. 
 A basal section was prepared, which in convergent polarized 
 light showed a normal, uniaxial interference figure, the double 
 refraction being negative, like the corresponding arsenic 
 compound. 
 
330 
 
 ON THE DOUBLE HALIDES OF 
 
 Fig. 2. 
 
 3RlBr.2SbBr*. Crystals 
 of this salt were prepared up 
 to 7mm. in diameter. The 
 habit is generally that of six- 
 sided plates, Fig. 2, having the 
 forms c, 0001, 0; r, 1011, 1 
 and 2, 0111, 1. In one crop 
 <?, r, and m were developed with 
 Fig. s. some of the rhombohedral faces 
 
 predominating to such an ex- 
 tent that the crystals looked like prisms, represented in basal 
 projection by Fig. 3. 
 
 r/\c, 10T1 A 0001 
 r A z, 10T1 A 01T1 
 
 Measured. 
 
 *52 21' 
 48 0' 
 
 Calculated. 
 
 47 57' 
 
 None of the crystals show normal optical properties. Crys- 
 tals like Fig. 3 showed for the most part an extinction parallel 
 to the direction c-d, sometimes with twinned lamellse promi- 
 nent at one end. In convergent polarized light no interference 
 figure was observed normal to c, but some of the crystals like 
 Fig. 3 could be tilted up on a rhombohedron face and showed 
 an acute bisectrix nearly normal to r. The axial angle was 
 small and the dispersion strong, the optical axes being in the 
 plane a b for green and normal to that for red light, the inter- 
 ference figure looking like that of brookite. 
 
 3RbI.2Sbls. This salt, unlike the corresponding arsenic 
 compound, was obtained in crystals of considerable size, some 
 over 10 mm. in diameter. When prepared with an excess of 
 SbI 8 , usually lozenge-shaped crystals were obtained, shown in 
 basal projection in Fig. 4. These were often grouped in twin 
 position, penetration twins being prominent with a rhombo- 
 hedron as twinning plane. From solutions containing an 
 excess of Rbl, the rhombohedral habit was observed, some- 
 times with a negative scalenohedron s 18.9.17, -^y-f, bevel- 
 ling its pole edges, Fig. 5. 
 
ANTIMONY WITH RUBIDIUM. 
 
 331 
 
 m A r, 10TO 
 r A c, 10T1 
 r A 3, 10T1 
 r A r, 10T1 
 
 S A 5, 18517 
 
 A 10T1 
 A 0001 
 
 A 0111 
 A 1101 
 
 A 19817 
 
 Measured. 
 
 *35 6' 
 
 54 43' 
 
 47 15' 
 
 89 50' 
 
 6 40' 
 
 Calculated. 
 48 18' 
 
 48 18' 
 
 90 14' 
 
 6 21' 
 
 The basal cleavage was prominent, and sections parallel to 
 this showed abnormal optical properties. Thin plates from 
 the crystals like Fig. 4 showed middle and end sections with 
 
 Fig. 4. 
 
 Fig. 5. 
 
 extinctions in the directions indicated by the arrows. In 
 convergent light no interference figure was observed. This 
 salt then, like 3RbBr.2SbBr 8 and 3RbI.2AsI 3 , is only pseudo- 
 hexagonal, being abnormal in its optical properties. 
 
 Rb CLSb C1 3 . The crystalli- 
 
 zation of this salt is monoclinic. 
 Crystals were obtained 10 mm. 
 long. The forms observed 
 were 
 
 Fig. 6. 
 
 a, 100, i-l 
 c, 001, 
 
 m, 110, / 
 d, 101, - 
 
 e, T01, 1-f 
 #111, 1 
 
 The habit is shown in Fig. 6. The axial ratio is d : b : c= 
 1.732 : 1.000 : 1.085 ; ft = 001 A 100 = 65 34'. 
 
332 
 
 ON THE DOUBLE HALIDES OF 
 
 a A c, 100 A 001 
 
 a A m, 100 A 110 
 
 c A e, 001 A T01 
 
 WA w, 110 A T10 
 
 a' A e, TOO A T01 
 
 p A e, TT1 A T01 
 
 p A c, TT1 A 001 
 
 c A d, 001 A 101 
 
 d A a, 101 A 100 
 
 Measured. 
 
 Calculated. 
 
 *65 34' 
 
 . . . 
 
 *57 37' 
 
 . . . 
 
 *37 36' 
 
 ... 
 
 64 46' 
 
 64 46' 
 
 76 54' 
 
 76 50' 
 
 46 28' 
 
 46 23' 
 
 56 50' 
 
 56 52' 
 
 24 21' 
 
 24 22' 
 
 41 24' 
 
 41 12' 
 
 772 
 
 Fig. 7. 
 
 In polarized light these crystals show an 
 extinction parallel to the ortho-axis. With 
 crystals flattened parallel to the basal plane 
 an obtuse bisectrix may be seen nearly 
 normal to the base, the plane of the optical 
 axis being at right angles to the symmetry 
 plane. 
 
 EbCU&S'bCli.H.fl. The crystallization of 
 this salt is monoclinic. Crystals were made 
 up to a length of about 9 or 10 mm. The 
 forms observed were : 
 
 a, 100, i-i 
 
 c, 001, 
 
 m, 110, I 
 
 d, 021, 2-^ 
 
 e, Oil, 14 
 
 p, 221, -2 
 a, T01, 1-* 
 
 The habit is shown in Fig. 7. The axial ratio is d : b : c 1.699 : 
 1 : 0.820 ; /? = 001 A 100 = 89 28 J'. 
 
 m*m, 110 A T10 
 a A a, 100 A T01 
 m A a, 110 A 100 
 a A c, 100 A 001 
 x A c, T01 A 001 
 c A e, 001 A Oil 
 c A d, 001 A 021 
 a A p, 100 A 221 
 p A d, 221 A 021 
 
 Measured. 
 
 Calculated. 
 
 *60 57' 
 
 . . 
 
 64 49' 
 
 64 40' 
 
 59 32' 
 
 59 32' 
 
 *89 28^' 
 
 . 
 
 *25 51' 
 
 . . . 
 
 39 10' 
 
 39 21' 
 
 58 20' 
 
 58 37' 
 
 63 5' 
 
 63 6' 
 
 27 5' 
 
 26 37' 
 
ANTIMONY WITH RUBIDIUM. 333 
 
 The crystals flattened parallel to the orthopinacoid show in 
 polarized light an extinction parallel to the ortho-axis, and in 
 convergent light a bisectrix and one of the ring systems appear 
 near the limits of the field. The plane of the optical axis is 
 the clinopinacoid. 
 
 The 23 : 10 bromide is pseudohexagonal. Basal plates 
 always showed an intricate twinning when examined in polar- 
 ized light. The pyramidal faces were horizontally striated 
 and to such an extent that no satisfactory measurements could 
 be made. In every respect this compound resembles the 
 23 : 10 chloride described by Remsen and Saunders. 
 
 In conclusion the author wishes to express his indebtedness 
 to Prof. H. L. Wells for valuable advice in connection with 
 the present investigation and to Prof. S. L. Penfield, under 
 whose generous supervision the crystallography of these salts 
 was investigated. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 June, 1893. 
 
ON THE DOUBLE CHLORIDES, BROMIDES, AND 
 IODIDES OF CAESIUM AND CADMIUM.* 
 
 BY H. L. WELLS AND P. T. WALDEN. 
 
 SINCE the caesium-mercuric halides f had been studied by 
 one of us with the result that six types of double salts were 
 found, it seemed desirable to extend the investigation to the 
 metal cadmium on account of its close relation to mercury. 
 We have, therefore, undertaken this work, and as the result of 
 a systematic and very thorough search have obtained the fol- 
 lowing compounds. The salt Cs 2 CdCl 4 had already been 
 described by Godeffroy. 
 
 3 : 1 Type. 2 : 1 Type. 1 : 1 Type. 
 
 .... Cs 2 CdCl 4 CsCdCl 8 
 
 Cs 3 CdBr 5 Cs 2 CdBr 4 CsCdBr 3 
 
 Cs 8 CdI 5 Cs 2 CdI 4 CsCdI 8 .H 2 
 
 These cadmium salts correspond to the three types of mercuric 
 compounds which contain the largest proportion of caesium, 
 and no evidence of the existence of cadmium double halides 
 analogous to the 2 : 3, 1 : 2, and 1 : 5 types of caesium-mercuric 
 salts could be obtained. It is evident that the tendency to 
 form a variety of double halides decreases from mercury to 
 cadmium. 
 
 Three types of cadmium double halides with alkali metals 
 and ammonium have been previously described, and a list of 
 these is as follows : 
 
 4 : 1 Type. 2 : 1 Type. 1 : 1 Type. 
 
 (NH 4 ) 4 CdCl 6 (NH 4 ) 2 CdCl 4 .H 2 KCdCl 3 4H 2 
 
 K 4 CdCl 6 Na 2 CdCl 4 .3H 2 NaCdBr 8 .2|H 2 O 
 
 (NH 4 ) 4 CdBr 6 K 2 CdCl 4 KCdBr 8 4H 2 
 
 K 4 CdBr e K 2 CdCl 4 .H 2 NH 4 CdBr 8 4H 2 O 
 
 (NH 4 ) 2 CdI 4 .2H 2 NH 4 CdF 8 
 
 Na 2 CdI 4 .6H 2 
 
 K 2 CdI 4 .2H 2 
 
 * Amer. Jour. Sci., xlvi, December, 1893. t Ibid., Ill, xliv, 221. 
 
CHLORIDES, ETC., OF CAESIUM AND CADMIUM. 335 
 
 It is noticeable that, while the 2 : 1 and 1 : 1 salts in the 
 above table correspond to two types of the caesium salts which 
 we have prepared, the 4 : 1 type of ammonium and potassium 
 compounds differs from our 3 : 1 caesium-cadmium salts and 
 from the corresponding caesium-mercuric compounds. We 
 were entirely confident that our results were correct, for the 
 salts were well crystallized and carefully prepared for analysis, 
 and it was impossible to believe that we had obtained too little 
 caesium in our analyses, because the salts of this type were 
 crystallized from solutions containing a large excess of caesium 
 halide. In order to convince ourselves that there was no mis- 
 take about the 4 : 1 formulas we have prepared the two chlo- 
 rides according to the directions of Von Hauer who described 
 them. The salts were extremely well crystallized and it was 
 easy to obtain them in a very pure condition. The results of 
 the analyses were as follows : 
 
 Potassium . 
 Cadmium . 
 Chlorine 
 
 Ammonium . . 
 Cadmium . . . 
 Chlorine . . . 
 
 These results confirm Von Hauer's formulae, and the curious 
 fact must be accepted that caesium forms 3 : 1 double halides 
 with cadmium, while potassium and ammonium form salts of 
 the 4 : 1 type. 
 
 The four types of cadmium double halides now known form 
 a very simple and symmetrical series, the ratios of the alkali 
 metal to cadmium being 4 : 1, 3 : 1, 2 : 1, and 1 : 1. The first 
 two of these types do not conform to Remsen's so-called law * 
 concerning the composition of double halides. 
 
 Preparation and General Properties. The compounds to 
 be described were prepared by making warm solutions of the 
 
 * Am. Chem. Jour., xi, 291. 
 
 Found. 
 
 Calculated for 
 K 4 CdCl. 
 
 32.35 
 
 32.49 
 
 23.39 23.36 
 
 23.27 
 
 44.00 44.12 
 
 44.24 
 
 Found. 
 
 Calculated for 
 (NH 4 ) 4 CdCl fl . 
 
 18.20 
 
 18.12 
 
 27.91 27.87 
 
 28.22 
 
 53.50 
 
 53.66 
 
336 DOUBLE CHLORIDES, BROMIDES, AND IODIDES 
 
 component halides, and after concentrating if necessary, cool- 
 ing to crystallization. Water, slightly acidified with the cor- 
 responding acid to prevent the formation of basic compounds, 
 was used as the solvent, and in one instance, where a solution 
 became syrupy from a large excess of a cadmium salt, alcohol 
 was also tried, but without any advantage. The conditions 
 were varied gradually in each case all the way from the point 
 where the solution was saturated with the caesium halide to 
 the point where it was saturated with the cadmium halide, 
 and so many experiments were made that we believe that no 
 double salt, capable of existence at the temperatures used, 
 was overlooked. It was noticed that variations in the concen- 
 tration of any given solution had little effect upon the identity 
 of the salt produced. In this respect the cadmium compounds 
 differ considerably from those of mercury, for with the latter 
 concentration is often an important factor in determining the 
 salt produced. 
 
 The three 1 : 1 compounds CsCdCls, CsCdBr 3 , and CsCdl,. 
 H 2 O and also the 2:1 iodide Cs 2 CdI 4 are capable of being 
 recrystallized from water unchanged. The salt Cs 2 CdCl 4 , 
 when dissolved in water, yields CsCdCl 3 , the two bromides 
 Cs 3 CdBr 6 and Cs 2 CdBr 4 yield CsCdBr 3 , while the iodide Cs 8 CdI 5 
 gives Cs 2 CdI 4 . These facts show that the salts having the 
 larger proportions of caesium require the presence of an excess 
 of caesium halide for their formation. The 1 : 1 salts all crys- 
 tallize unchanged from extremely concentrated solutions of 
 the corresponding cadmium halides. 
 
 All the salts are colorless. A pale violet color noticed in a 
 few crops of the bromide Cs 2 CdBr 4 is supposed to have been 
 due to some unknown foreign substance. 
 
 The solubility of the analogous salts in water or in saline 
 solutions evidently increases from the chlorides to the iodides. 
 The iodides consequently yield the largest crystals, while the 
 chlorides give the smallest. 
 
 Methods of Analysis. The products were carefully exam- 
 ined, and nothing was analyzed that was not homogeneous. 
 The crystals, which, in several instances, were large and fine 
 
OF CESIUM AND CADMIUM. 337 
 
 and in no case hygroscopic, were freed from mother-liquor 
 with great care by pressing and crushing them on smooth 
 filter-paper. They were then simply air-dried for analysis. 
 
 Cadmium was precipitated as sulphide, this was dissolved 
 in hydrochloric acid containing bromine, and after the free acid 
 had been removed by evaporation, the cadmium was precipi- 
 tated with potassium carbonate solution, and cadmium oxide 
 was weighed on a Gooch filter. The caesium in the filtrate 
 from the cadmium sulphide was determined as normal sulphate. 
 The halogens were determined in separate portions by the 
 usual gravimetric method. 
 
 In every case at least two separate crops of a salt were 
 made and analyzed, so as to avoid any chance of mistakes 
 arising from mixtures. 
 
 2:1 C cesium- Cadmium Chloride, Cs 2 CdCl^. This salt is 
 produced as a precipitate when a solution of cadmium chloride 
 is added to a concentrated caesium chloride solution. The 
 precipitate dissolves upon warming the liquid, and crystallizes 
 out in very small, rectangular plates when the solution is 
 cooled. Its formation was observed when 50 g. of caesium 
 chloride and 3 g. of cadmium chloride were used, and it con- 
 tinued to be produced with the same amount of caesium chloride 
 until the amount of cadmium chloride had been increased to 
 18 g., at which point the 1 : 1 salt began to form. The salt 
 is very sparingly soluble in caesium chloride solutions, and it 
 is probably due to this fact that no chloride of the 3 : 1 type 
 could be obtained. 
 
 Three separate crops gave the following results on analysis : 
 
 Found. Calculated for 
 
 Caesium . . . 51.55 51.26 51.51 51.35 
 
 Cadmium . . 21.45 21.50 . . . 21.62 
 
 Chlorine . . 27.03 27.14 26.90 27.03 
 
 100.03 99.90 100.00 
 
 1:1 Ccesium- Cadmium Chloride, CsCdCl y This was ob- 
 tained only as a white crystalline powder. It is formed under 
 
 22 
 
338 DOUBLE CHLORIDES, BROMIDES, AND IODIDES 
 
 a very wide range of conditions, being produced by the re- 
 crystallization of the preceding salt, and continuing to appear 
 until the solution is saturated with cadmium chloride. It is 
 very difficultly soluble, especially in concentrated cadmium 
 chloride solutions, and it can be recrystallized unaltered from 
 water. Two samples, obtained under widely different condi- 
 tions, were analyzed. 
 
 Caesium .... 38.11 37.60 37.84 
 
 Cadmium . . . 31.80 31.97 31.86 
 
 Chlorine . . . 30.17 30.25 30.30 
 
 100.08 99.82 100.00 
 
 3:1 Ccesium- Cadmium Bromide, Os 3 CdBr^. This com- 
 pound was obtained in the form of rectangular plates, some- 
 times as much as 10 mm. in diameter. It can be made from a 
 solution of 80 g. of caesium bromide and 4.5 g. of cadmium 
 bromide in sufficient water to make a volume of 120 c. c. On 
 recrystallization it gives CsCdBr 3 . 
 
 Two separate samples were analyzed. 
 
 Found. 
 
 Cs 3 CdBr 5 
 
 Caesium . . . 44.25 44.27 43.80 
 
 Cadmium ... 11.88 . . . 12.29 
 
 Bromine . . . 43.79 43.77 43.91 
 
 99.92 100.00 
 
 2:1 Caesium- Cadmium Bromide, Cs^CdBr^. This was 
 obtained in the form of slender needles, usually colorless, but 
 sometimes possessing a pale violet color for some unknown 
 reason. A solution 120c. c. in volume, containing 3g. of 
 cadmium bromide and 52 g. of caesium bromide gave this salt. 
 When recrystallized from water, it gives, like the preceding 
 salt, the compound CsCdBr s . 
 
 The following analyses of separate crops were made. No. 
 IV was a simple of the pale violet variety. 
 
OF CESIUM AND CADMIUM. 339 
 
 _ _ ^ Calculated for 
 
 T II. III. IV. ^ Cs 2 CdBr 4 . 
 
 Cs . . . 40.46 40.53 . . . 40.46 38.11 
 
 Cd . . . 14.55 14.62 14.68 14.78 16.05 
 
 Br . . . 45.12 44.97 45.04 45.04 45.84 
 
 100.13 100.12 100.28 100.00 
 
 Although the analyses agree well among themselves, it is 
 noticeable that they vary considerably from the calculated 
 composition. This disagreement is probably due to contamina- 
 tion with caesium chloride, resulting from the large surface 
 exposed by the slender crystals and the concentration of the 
 mother-liquor. Moreover, analogy with the chloride and iodide 
 makes the simple formula Cs 2 CdBr 4 far more probable than the 
 complicated formula Cs 7 Cd 8 Br 18 with which the analyses 
 correspond. 
 
 1:1 Ccesium- Cadmium Bromide, Cs CdBr s . The condi- 
 tions under which this compound is formed are very wide in 
 range, for it is produced by recrystallizing Cs 8 CdBr 6 , and it 
 continues to appear as cadmium bromide is added until the 
 solution is saturated with this very soluble salt. 
 
 The compound is evidently dimorphous. One form occurs 
 as a crystalline precipitate, apparently isometric in form, under 
 narrow limits of conditions when caesium bromide is in excess, 
 being produced when Cs 3 CdBr 6 is recrystallized from water. 
 The other form occurs in well-crystallized prisms, and is 
 obtained when Cs 2 CdBr 4 is recrystallized and when cadmium 
 bromide is in excess of this proportion. It is interesting to 
 notice that we have described a caesium-lead bromide * of this 
 type, CsPbBr 8 , which is dimorphous, and that one of us has 
 described the dimorphous mercuric compounds,! CsHgCl 3 and 
 CsHgBr 3 , which also belong to the same type. 
 
 Below are the analyses of four separate samples. Number 
 IV is the granular, isometric salt ; the others represent the 
 prismatic compound. 
 
 * Amer. Jour. Sci., xlv, 128. 
 t Ibid., xliv, 225 and 228. 
 
340 DOUBLE CHLORIDES, BROMIDES, AND IODIDES 
 
 Calculated for 
 
 34.85 
 
 34.89 
 
 34.78 
 
 
 vqgVNUfi 
 
 34.82 
 
 9.78 
 
 9.79 
 
 9.84 
 
 9.55 
 
 9.77 
 
 55.23 
 
 55.35 
 
 55.32 
 
 55.36 
 
 55.41 
 
 t _ _ ^ 
 
 T U. III. IV? CsCdBr 3 . 
 
 Cs . . . 27.67 27.48 . . . 27.95 27.42 
 
 Cd . . . 22.97 23.08 22.87 22.92 23.09 
 
 Br . . . 49.49 49.42 49.33 49.30 49.49 
 
 100.13 99.98 100.17 100.00 
 
 3 : 1 Ccesium- Cadmium Iodide, Cs s CdI 5 . This salt crystal- 
 lizes beautifully in large, stout, twinned prisms which show a 
 variety of habits. Some of the crystals obtained were as 
 much as 50mm. in diameter. Its formation was observed 
 when 182 g. of caesium iodide and 65 g. of cadmium iodide were 
 dissolved in sufficient water to make a volume of 200 c. c. 
 
 Four crops gave the following results on analysis : 
 
 Found. Calculated for 
 
 Cs . . . 
 
 Cd . . . 
 I ... 
 
 99.86 100.03 99.94 100.00 
 
 2:1 Ccesium- Cadmium Iodide, C8 2 CdI. This, like the 
 corresponding mercuric salt, crystallizes in nearly square plates, 
 in prisms, and in intermediate forms. Some of the plates 
 obtained were 50 or 75 mm. in diameter. It can be prepared 
 by recrystallizing Cs 8 CdI 5 from water, and as the proportion 
 of cadmium iodide is increased, it continues to form until the 
 ratio of cadmium to caesium has almost reached 1:1. The 
 range of its formation is, therefore, much greater than that of 
 the corresponding chloride and bromide, and it also differs 
 from these in being recrystallizable from water. Three different 
 samples were analyzed. 
 
 Caesium . 
 Cadmium 
 Iodine 
 
 100.09 100.11 100.00 
 
 1:1 Ccesium-Cadmium Iodide, CsCdT s .ff z O. This salt 
 forms thin plates, often 20 to 30 mm. in diameter. It is the 
 
 
 Found. 
 
 
 Calculated for 
 
 Cs 3 CdI 4 . 
 
 30.03 
 
 29.85 
 
 30.29 
 
 30.23 
 
 12.56 
 
 12.53 
 
 12.46 
 
 12.64 
 
 . . . 
 
 57.27 
 
 57.42 
 
 57.33 
 
OF CAESIUM AND CADMIUM. 341 
 
 only hydrous caesium-cadmium halide that we have obtained, 
 and it is stable when exposed to the air at ordinary temperatures. 
 
 It was considered doubtful whether the corresponding 
 caesium mercuric iodide * contained a molecule of feebly com- 
 bined water or not, but since both the cadmium and mercuric 
 salts crystallize in thin plates, it is now believed, from analogy, 
 that the mercuric compound was really hydrous. 
 
 The compound under consideration is formed under wide 
 limits of conditions when the cadmium present is atomically 
 equivalent to or in excess of the caesium. It can be recrystal- 
 lized from water. The samples analyzed were prepared under 
 widely varying conditions. 
 
 Pound. Calculated for 
 
 , ' ^ C8CdI 8 .H 2 0. 
 
 Caesium .... 20.89 20.75 . . . 20.66 
 
 Cadmium . . . 17.13 17.43 17.89 17.39 
 
 Iodine .... 59.21 59.18 . . . 59.16 
 
 Water .... 2.88 2.80 2.76 2.79 
 
 100.11 100.16 100.00 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 August, 1893. 
 
 * Amer. Jour. Sci., xliv, 230. 
 
ON THE DOUBLE CHLORIDES, BROMIDES, AND 
 
 IODIDES OF CAESIUM AND ZINC, AND OF 
 
 CESIUM AND MAGNESIUM.* 
 
 BY H. L. WELLS AND G. F. CAMPBELL. 
 
 THE caesium-mercuric and the caesium-cadmium halides have 
 already been studied in this laboratory, and it has seemed 
 desirable to extend the investigation to the zinc and magne- 
 sium compounds, thus completing the study of the caesium 
 double halides of this family of bivalent metals as far as the 
 chlorides, bromides, and iodides are concerned. 
 We have obtained the following salts ; 
 
 3 : 1 Type. 2 : 1 Type. 1 : 1 Type. 
 
 Cs 3 ZnCl 5 Cs 2 ZnCl 4 ? 
 
 Cs 3 ZnBr 5 Cs 2 ZnBr 4 ? 
 
 Cs 3 ZnI 6 Cs 2 ZnI 4 ? 
 
 .... .... CsMgCl 3 .6H 2 
 
 .... .... CsMgBr 3 .6H 2 
 
 A systematic and thorough search has been made in all cases, 
 and it is remarkable that while mercury gave six types of 
 caesium double salts and cadmium gave three, only two could 
 be obtained with zinc and one with magnesium. It is evident 
 that the variety of these double salts increases with the atomic 
 weight of the bivalent metal. The existence of zinc salts of 
 the 1 : 1 type is suspected, but the suspected products were 
 obtained only in extremely concentrated zinc halide solutions 
 of such a syrupy nature that no satisfactory analyses of them 
 could be made. 
 
 The previously described double halides of zinc and mag- 
 nesium with the alkali metals, as far as we have been able to 
 find them, are given in the following table : 
 
 * Amer. Jour. Sci., xlvi, December, 1893. 
 
CHLORIDES, ETC., OF CESIUM AND ZINC. 343 
 
 3 : 1 Type. 2 : 1 Type. 1 : 1 Type. 
 
 (NH 4 ) 3 ZnCl 6 (NH 4 ) 2 ZnCl 4 NH 4 ZnCl 8 .2H 8 
 
 (NH 4 ) 2 ZnCl 4 .H 2 KZnI 8 
 
 Na 2 ZnCl 4 .3H 2 NaZnI 8 .liH 2 
 
 K 2 ZnCl 4 NaZnF 8 
 
 (NH 4 ) 2 ZnBr 4 KZnF 8 
 
 (NH 4 ) 2 ZnI 4 NH 4 MgCl 8 .6H 2 
 
 Na 2 ZnI 4 .3H 2 NaMgCl 8 .H 2 O 
 
 K 2 ZnI 4 KMgCl 8 .6H 2 
 
 K 2 ZnF 4 RbMgCl 3 .6H 2 O 
 
 (NH 4 ) 2 ZnF 4 .2H 2 NH 4 MgBr 8 .6H 
 
 KMgBr 8 .6H 2 
 
 NH 4 MgI 3 .6H 2 
 
 KMgI 3 .6H 2 
 
 NaMgF 3 
 
 There is but a single 3 : 1 salt, corresponding to our new 
 caesium compounds of that type. This was described by 
 Marignac. A few 1 : 1 zinc salts have been described, hence 
 it is remarkable that caesium zinc salts of this type could not 
 be obtained in a pure condition, for previous experience in this 
 laboratory has shown that caesium usually forms less soluble 
 and more stable double halides than the other alkali metals. 
 All the previously described magnesium salts belong to the 
 1 : 1 type* to which our caesium salts belong, and like the 
 latter nearly all have six molecules of water. 
 
 The caesium-magnesium bromide is formed under narrower 
 limits of conditions than the chloride, while no iodide could 
 be prepared, for caesium iodide crystallized unchanged from 
 syrupy solutions of magnesium iodide. This behavior was 
 quite unexpected in view of the fact that the ammonium and 
 potassium double iodides are known, and we have here another 
 instance where caesium, in spite of its usual tendency to form 
 double salts, is inferior in this respect to other alkali metals. 
 The idea suggests itself that great differences between the 
 
 * Lerch has shown (J. Pr. Ch., II, xxviii, 338) that the salts 2KBr.MgBr 2 . 
 6H 2 O and 2NH 4 Br.MgBr 2 .6H 2 of Lowig do not exist. 
 
DOUBLE CHLORIDES, BROMIDES, AND IODIDES 
 
 atomic weights of the alkali-metal and the less positive metal 
 are unfavorable for the formation of double salts, but more 
 facts will be necessary in order to establish such a rule. 
 
 The caesium-magnesium salts show an increase in ease of 
 formation from the iodide to the chloride. Such a gradation, 
 both in variety of salts and ease of preparation, is evident in 
 a number of series of double halides which have been studied 
 in this laboratory, and the well-known tendency of fluorides 
 to form double salts indicates that the gradation probably 
 extends to these compounds. 
 
 3 : 1 C cesium- Zinc Chloride, Bromide, and Iodide, Cs 8 ZnCl 6 , 
 Cs z ZnBr & , and Cs s ZnI 6 . Each of these salts crystallizes in 
 colorless prisms, apparently monoclinic in form. They are 
 produced by making aqueous solutions of the constituents in 
 the calculated proportions, but in the case of the iodide, with 
 these proportions the 2 : 1 salt may form if the solution is too 
 dilute. The salts under consideration continue to form as the 
 relative amounts of caesium halides are increased until the 
 latter crystallize upon them. This indicates that no double 
 salts with a higher proportion of caesium exist. The iodide 
 forms under rather wider limits of conditions than the other 
 two salts, and it usually forms larger crystals. All the salts 
 require concentrated caesium halide solutions for their prepara- 
 tion, and the chloride especially is difficult to obtain in suffi- 
 cient quantity for analysis unless as much as one or two 
 hundred grams of the caesium halide is used. The following 
 analyses were made, all of which represent separate crops : 
 
 Found. 
 _ 
 
 Csesium ....... 62.46 . . . 62.20 
 
 Zinc ........ 10.08 9.80 10.13 
 
 Chlorine ....... 27.43 27.34 27.67 
 
 Csesium . 
 Zinc . . 
 Bromine 
 
 
 Pound. 
 
 
 Calculated for 
 Cs 3 ZnBr fl . 
 
 
 
 
 47.12 
 
 . . . 
 
 . 
 
 46.18 
 
 7.32 
 
 7.54 
 
 7.87 
 
 7.52 
 
 45.91 
 
 46.54 
 
 45.85 
 
 46.30 
 
OF CAESIUM AND ZINC, ETC. 345 
 
 Found. Calculated for 
 x Cs 8 ZnI 8 . 
 
 Caesium. . . . 36.54 36.20 36.08 36.30 
 
 Zinc ..... 5.77 5.95 5.74 5.92 
 
 Iodine .... 56.89 57.16 . . . 57.78 
 
 2 : 1 Ccesium-Zinc Chloride, Bromide and Iodide, Cs 2 ZnCl 4 , 
 Cs^ZnBr^ and Cs^Znl^. These salts form colorless plates, de- 
 creasing in size from the iodide to the chloride. They are all 
 readily produced when larger proportions of the zinc halides 
 are used than are necessary for the 3 : 1 compounds, and they 
 recrystallize from water unchanged. They continue to form, 
 through a wide range of conditions, as more of the zinc 
 halides are added until the solutions become syrupy. In 
 extremely syrupy solutions crystals of a different appearance 
 were noticed, but on account of the nature of these solutions, 
 no satisfactory analyses of these products could be made. It 
 seems probable that they were 1 : 1 salts, analogous to the 
 cadmium salts of that type. 
 
 The following analyses of separate crops were made : 
 
 -, Calculated for 
 
 Cs 2 ZnCl 4 . 
 
 Caesium ....... 55.97 56.09 56.26 
 
 Zinc ........ 13.49 13.87 13.72 
 
 Chlorine ....... 29.89 29.97 30.02 
 
 Calculated for 
 Cs 2 ZnBr 4 
 
 Caesium ....... 40.68 . . . 40.86 
 
 Zinc ........ 9.53 9.72 9.98 
 
 Bromine ....... 49.30 49.17 49.16 
 
 Calculated for 
 Found - Cs,ZnI 4 . 
 
 Caesium ....... 31.49 31.55 31.70 
 
 Zinc ........ 7.61 7.82 7.75 
 
 Iodine ........ 60.43 . . . 60.55 
 
 Ccesium-Magnesium Chloride and Bromide, 
 and CsMgBr^.GH^O. These salts form colorless, rectangular 
 plates or flat prisms which are often striated. A thorough 
 search gave no indications of salts of other types. The 
 
346 CHLORIDES, ETC., OF CAESIUM AND ZINC. 
 
 chloride is formed under a wide range of conditions, the 
 bromide under a much narrower range, while no double iodide 
 at all could be prepared. 
 
 The following analyses were made of separate products : 
 
 Found. 
 
 Calculated for 
 
 Cs . 
 
 Mg. 
 Cl . 
 H 8 
 
 37.14 
 680 
 
 . . . 
 
 35.66 
 
 683 
 
 35.77 
 653 
 
 29.84 
 
 30.13 
 30.93 
 
 29.70 
 
 29.55 
 
 28.65 
 29.05 
 
 Found. 
 
 Cs . . . 
 
 . . 27.23 
 
 27.67 
 
 ~"* 
 
 Mg . 
 
 . . 4.96 
 
 4.50 
 
 5.07 
 
 Br . . . 
 
 . . 48.93 
 
 48.65 
 
 
 H 2 . . 
 
 . . 18.32 
 
 
 22.33 
 
 Calculated for 
 CsMgBr s .6H 2 0. 
 
 26.32 
 
 4.81 
 
 47.51 
 
 21.37 
 
 It should be mentioned that Dr. H. L. Wheeler of this 
 laboratory has attempted to prepare a double chloride of 
 caesium and beryllium. He found -that the simple salts 
 crystallized side by side from sufficiently concentrated solu- 
 tions, and there were no indications of the existence of any 
 double salt, even at rather low temperatures. It is therefore 
 evident that beryllium follows the rule, already indicated, that 
 in this family of bivalent elements, Be, Mg, Zn, Cd, Hg, the 
 tendency to form double halogen salts increases with their 
 atomic weights. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 August, 1893. 
 
ON THE C^ESIUM-CUPRIC CHLORIDES.* 
 
 BY H. L. WELLS AND L. C. DUPEE. 
 
 As a continuation of the work done in this laboratory on 
 double halogen salts, we have taken up the csesium-cupric 
 chlorides, which had never been thoroughly investigated. The 
 result has been the discovery of four double salts belonging to 
 three different types. The beauty of the crystals in size and 
 form, and the magnificent and unexpected colors of some of 
 them have made the investigation a very interesting one. The 
 colors of the anhydrous salts, yellow and red, were perhaps not 
 very remarkable since anhydrous cupric chloride is reddish 
 brown, but since water of crystallization is supposed to give 
 green and blue colors to cupric salts, we were considerably 
 surprised to find that a brown salt, CssCuaCl^HaO, was 
 hydrous. The color of this hydrous salt is, however, not 
 without analogy, for a garnet-red, hydrous lithium-cupric 
 chloride is known, LiCuCl 2 .2|H 2 O according to Chassevant,f 
 or LiCuCl 8 .2H 2 O according to Meyerhoffer ; $ moreover Engel 
 has described a garnet-red compound HCuCl 3 .3H 2 O, and 
 Sabatier's red salt H 2 CuCl 4 .5H 2 O, || is similarly exceptional 
 in color. 
 
 In this connection it should be noticed that cupric chloride, 
 when dissolved in water with an excess of caesium chloride, 
 gives a bright yellow solution when it is hot and concentrated. 
 It is well known that solutions of cupric chloride in concen- 
 trated hydrochloric acid have the same yellow color. 
 
 A list of the formulas of the salts to be described, with their 
 colors, is given below. The first salt has already been de- 
 scribed by Godeffroy.^]" 
 
 * Amer. Jour. Sci., xlvii, Feb. 1894. t Compt. rend., cxiii, 646. 
 
 J Monatshefte, xiii, 716. Compt. rend., cvi, 273. 
 
 || Ibid., cvi, 1724. T Berichte, viii, 9. 
 
348 THE C^ESIUM-CUPRIC CHLORIDES. 
 
 Cs 2 CuCl 4 Brilliant yellow. 
 
 Cs 2 CuCl 4 .2H 2 O Bluish green. 
 
 Cs 3 Cu 2 Cl r 2H 2 Brown. 
 
 CsCuClg Garnet-red. 
 
 The previously described cupric double halides containing 
 alkali metals and ammonium belong to two of the types which 
 we have found in investigating the caesium-cupric chlorides. 
 A list of all those that we have been able to find is given 
 below. Four of the double fluorides have been recently de- 
 scribed by Von Helmont.* 
 
 2 : 1 Type. 1 : 1 Type. 
 
 (NH 4 ) 2 CuCl 4 .2H 2 NH 4 CuCl 3 .2H 2 
 K 2 CuCl 4 .2H 2 NH 4 CuF 3 .2H 2 
 K 2 CuF 4 KCuF 8 
 (NH 4 ) 2 CuF 4 .2H 2 RbCuF 8 
 LiCuCl 8 .2H 2 
 
 It is to be noticed that this list contains salts which corre- 
 spond exactly to three of the caesium compounds, and that the 
 group of 2 : 1 salts with two molecules of water is a conspicu- 
 ous one. 
 
 The salt Cs 3 Cu 2 Cl 7 .2H 2 O is an interesting one because it 
 is apparently the only known double halide of an alkali metal 
 with a bivalent metal, which has the 3 : 2 ratio. 
 
 The caesium salts were investigated systematically by start- 
 ing with a solution of 50 g. of caesium chloride and adding to 
 this from 3 to 5 g. of cupric chloride at a time, evaporating 
 after each addition and observing the products. At the same 
 time another series of experiments was made by beginning 
 with a solution of 50 g. of cupric chloride, adding caesium 
 chloride to this gradually and operating in the same manner 
 as in the other case. Many additional experiments were made, 
 sometimes with the use of as much as 200 g. of caesium chlo- 
 ride, and a number of crystallizations were made in the pres- 
 ence of hydrochloric acid of various strengths. It is believed 
 
 * Zeitschr. anorg. Chem., iii, 115. 
 
THE CJZSIUM-CUPRIC CHLORIDES. 349 
 
 that no double salt capable of existence either in warm solu- 
 tions or at ordinary temperatures has been overlooked. 
 
 The salts were so well crystallized and so distinct in form 
 and color that there was no difficulty in selecting pure products 
 for analysis. The usual precautions, often mentioned in com- 
 munications from this laboratory, were taken for the removal 
 of mother-liquor from the crystals. 
 
 In analyzing the salts copper and caesium were determined 
 in one portion, the first as subsulphide, the other as normal 
 sulphate. The chlorine was determined in separate portions, 
 by the usual gravimetric method. 
 
 Anhydrous 2 : 1 Ccesiurrir Cupric Chloride, Cs 2 CuCl 2 . 
 This salt, which Godeffroy first described, forms magnificent, 
 yellow, orthorhombic prisms, which were often obtained sev- 
 eral centimeters in length and several millimeters in thickness. 
 The crystals are usually attached at one end, and they often 
 arrange themselves in parallel position, forming flat clusters. 
 Doubly terminated, short crystals were occasionally observed. 
 Its formation was observed, with 50 g. of caesium chloride, in 
 the presence of from 5 to about 25 g. of cupric chloride. It 
 can be recrystallized from water if the solution is made so con- 
 centrated that crystals form on cooling, but with more dilute 
 solutions one or both of the hydrous salts are usually deposited 
 on standing or on spontaneous evaporation. The following 
 analyses represent different crops made under considerable 
 differences of conditions : 
 
 Found. Calculated for 
 
 Cs 2 CuCl4. 
 
 56.42 
 13.46 
 30.12 
 100.00 
 
 Cs . 
 
 r 
 
 56.33 
 
 56.14 
 
 56.18 
 
 Cu . 
 Cl . 
 
 . . 13.52 
 . . 30.07 
 
 13.45 
 29.99 
 99.77 
 
 13.47 
 30.04 
 99.65 
 
 13.48 
 30.03 
 99.69 
 
 Hydrous 2 : 1 Ccesium- Cupric Chloride, 
 This salt is bluish green in color, and it loses its water very 
 rapidly on exposure to the air with a change of color to bright 
 yellow. It is a well-crystallized, transparent salt, but its form 
 was not made out on account of its instability. It is difficult 
 
350 THE CJESIUM-CUPRIC CHLORIDES. 
 
 to prepare it, at least at summer temperatures under which 
 this investigation has been made, and we have only occasion- 
 ally observed it. It is formed by allowing solutions contain- 
 ing nearly the required proportions of csesium and copper 
 chlorides to evaporate spontaneously. A sample quickly 
 pressed on paper gave the following analysis : 
 
 ,. . Calculated for 
 
 Found. Cs 2 CuCl 4 .2H 2 0. 
 
 Caesium 51.28 52.40 
 
 Copper 12.53 12.50 
 
 Chlorine 28.00 
 
 Water 7.20 7.10 
 
 Another sample, which had been exposed to the air too long, 
 gave 6.02 per cent of water, and the dehydrated compound 
 gave the following analysis : 
 
 Found. Calculated. 
 
 Cesium 56.09 56.42 
 
 Copper 13.68 13.46 
 
 3 : 2 Ccesium- Cupric Chloride, Os 3 Ou 2 C1 7 .2R 2 0. This 
 compound was obtained from solutions of nearly the required 
 proportions of csesium and cupric chlorides. It usually forms 
 only at ordinary temperatures, and if the solution is too con- 
 centrated, one or both of the anhydrous salts will be deposited 
 while it is warm. The salt forms triclinic crystals, often one 
 or two centimeters in diameter The large crystals are deep 
 brown in color, small ones and fragments are very much paler, 
 while the powder is yellow. It is nearly stable at ordinary 
 temperatures, but gradually loses its lustre on long exposure. 
 All the water goes off readily at 100. The following analyses 
 of separate crops were made : 
 
 Caesium 
 Copper 
 Chlorine . 
 Water . . 
 
 " 99.54 99.80 100.00 
 
 
 Found. 
 
 
 Calculated for 
 Cs 8 Cu 2 Cl 7 .2H 2 O. 
 
 49.23 
 
 . . . 
 
 49.36 
 
 48.96 
 
 15.68 
 
 15.90 
 
 15.74 
 
 15.67 
 
 30.84 
 
 29.90 
 
 30.69 
 
 30.66 
 
 4.22 
 
 4.38 
 
 4.41 
 
 4.44 
 
THE C^ESIUM-CUPRIC CHLORIDES. 351 
 
 1:1 C cesium- Cupric Chloride, CsCuCl z . This is formed 
 under wide variations of conditions, up to the point where the 
 solution is saturated with cupric chloride. It can be recrystal- 
 lized from water. It forms slender hexagonal prisms termi- 
 nated by pyramids. The color is a deep garnet-red, and all 
 except very slender crystals appear black by reflected light. 
 The following analyses of separate products were made : 
 
 Found. Calculated for 
 . CsCuCl,. 
 
 Caesium . . . 43.67 . . . 43.58 43.89 
 
 Copper . . . 21.16 21.17 21.06 20.96 
 
 Chlorine . . 35.25 35.22 35.00 35.15 
 
 100.08 99.64 100.00 
 
 FIELD SCIENTIFIC SCHOOL, 
 September, 1893. 
 
ON THE C^ESIUM-CUPRIC BROMIDES.* 
 
 BY H. L. WELLS AND P. T. WALDEN. 
 
 WE have made a systematic investigation of the caesium- 
 cupric bromides, following the plan, described in the preced- 
 ing article, which was used for the corresponding chlorides. 
 Although the work has been very thorough, we have found 
 only the following two salts : 
 
 Cs 2 CuBr 4 and CsCuBr 8 . 
 
 These salts correspond to the two common types of cupric 
 double halides. The fact that no hydrous salts could be 
 obtained was unexpected, because it has been pointed out by 
 Remsen f in the case of certain double halides, and it has been 
 observed by one of us in the case of the alkaline-lead halides, J 
 that the tendency to combine with water seems to increase 
 with the atomic weight of the halogen. The fact that hydrous 
 double chlorides of caesium and copper exist, while no corre- 
 sponding bromides were obtained indicates that the rule does 
 not apply in all cases. 
 
 2:1 Caesium- Cupric Bromide, Cs^CuBr^ This compound 
 forms opaque, black crystals having a greenish tint. The 
 powder is black. In form and habit it resembles the corre- 
 sponding chloride and is evidently orthorhombic like that salt. 
 Elongated prisms, usually not over 5 to 10 mm. in length, 
 commonly occurring in groups in parallel position, were 
 observed where an excess of caesium bromide was used. 
 When the proportion of cupric bromide was increased, small 
 short crystals made their appearance. 
 
 With 50 g. of caesium bromide the compound is formed in 
 the presence of from 5 to about 70 g. of cupric bromide. The 
 
 * Amer. Jour. Sci., xlvii, February, 1894. 
 
 t Amer. Chem. Jour., xiv, 88. } Amer. Jour. Sci., xlvi, 37. 
 
THE C^ESIUM-CUPRIC BROMIDES. 353 
 
 range of conditions under which it is formed are considerably 
 wider than in the case of the corresponding chloride. 
 
 Of the analyses given below, A, B, C, D, and E represent a 
 series of preparations in which copper bromide was gradually 
 increased from 5 g. in A to 46 g. in E, while the caesium bro- 
 mide remained constant at about 50 g. and the volume increased 
 from 100 c. c. in A to 150 c. c. hi E. The sample F was obtained 
 by recrystallizing the salt from water, while G resulted from 
 recrystallizing CsCuBr 8 . 
 
 Cs. Cu. Br. 
 
 A .... 40.80 . 9.46 49.38= 99.64 
 
 B . . 9.68 
 
 C . . . . 40.75 9.74 48.97= 99.46 
 
 D . . . . 40.88 9.69 48.95= 99.52 
 
 E ........ 9.78 49.40 
 
 F . . . . 41.11 9.66 
 
 G 10.00 
 
 Calculated j 4927-10000 
 
 forCs 2 CuBrJ 4 
 
 1:1 Ccesium- Cupric Bromide Os CuBr z . This salt forms 
 short, hexagonal crystals which are strung together end to 
 end. They are dark and opaque, giving a bronze-colored 
 reflection, while their powder is nearly black. When recrys- 
 tallized from water, the compound gives the 2:1 salt, thus 
 differing from the chloride. It was obtained from a solution 
 containing 50 g. of caesium bromide and 70 g. of copper bromide 
 with sufficient water to form a volume of 200 c.c., and it con- 
 tinued to be produced as cupric bromide was added until the 
 solution became saturated with that compound. 
 
 The following analyses were made of separate preparations 
 which were obtained under wide variations of conditions. 
 
 Cs . . 
 Cu . . . 
 Br . . . 
 
 99.74 99.14 100.20 100.00 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 September, 1893. 
 
 23 
 
 Found. 
 
 Calculated 
 for CsCuBr 
 
 29.93 
 
 29.09 
 
 . . . 30.43 
 
 30.48 
 
 14.72 
 
 15.09 
 
 14.73 14.81 
 
 14.53 
 
 55.09 
 
 54.96 
 
 . . . 64.96 
 
 54.99 
 
ON THE CAESIUM-CUPROUS CHLORIDES.* 
 
 BY H. L. WELLS. 
 
 THE salts to be described were prepared by heating solutions 
 containing caesium chloride and cupric chloride with copper 
 wire, and sufficient hydrochloric acid to prevent the formation 
 of basic salts ; then, after the copper in solution was chiefly in 
 the form of cuprous chloride, cooling to crystallization. 
 
 When the solutions were dilute, caesium chloride being in 
 excess, very slender white prisms were obtained under wide 
 variations of conditions. The crystals became yellowish while 
 being dried with paper, but they were apparently nearly stable 
 in the air when dry. It was found that the salt was decom- 
 posed by water. Two different products were analyzed. 
 
 ip nnn i Calculated for 
 
 ound> CsCl.Cu z Cl 2 . 
 
 Caesium 36.93 36.36 36.29 
 
 Copper 34.33 34.17 34.64 
 
 Chlorine .... 28.95 28.87 29.07 
 
 The results show that the formula CsCu 2 Cl 3 belongs to this 
 salt. 
 
 On using more concentrated solutions, also with an excess 
 of caesium chloride, thin, rectangular, colorless plates were 
 produced, sometimes 10 or 20 mm. in diameter. The range of 
 conditions under which this salt is produced is wide, and large 
 crops of it are easily prepared. As the concentration of the 
 caesium chloride solutions was increased, the same compound 
 appeared in the form of blade-like crystals with pointed ends. 
 By dissolving this salt in water the previously described com- 
 pound is produced by crystallization. The surface of the 
 crystals becomes yellow on drying, but when dry it appears to 
 
 * Amer. Jour. Sci., xlvii, February, 1894. 
 
THE CAESIUM-CUPROUS CHLORIDES. 355 
 
 be very stable. The first two analyses represent separate 
 crops of the rectangular plates, the third a crop of the blade- 
 like crystals. 
 
 Found. 
 
 Caesium 
 Copper . 
 
 i. 
 .... 56.81 
 .... 17.95 
 
 n. 
 56.66 
 1789 
 
 m. 
 56.84 
 17.84 
 
 uaicuiatea xor 
 SCsCl.CujClj. 
 
 56.72 
 1805 
 
 Chlorine 
 
 .... 25.03 
 
 25.08 
 
 25.13 
 
 25.23 
 
 It is evident that this salt has the formula Cs 8 Cu 2 Cl 6 . 
 
 With nearly or quite saturated caesium chloride solutions 
 containing comparatively little cuprous chloride, prismatic 
 crystals are formed on cooling. The crystals are very pale 
 yellow in color, and their lustre is less brilliant than that of 
 the preceding compound. Crystals having a diameter of two 
 or three millimeters and a length of several centimeters were 
 sometimes observed. This salt forms under very narrow limits 
 of conditions, and it is very difficult to obtain it free from the 
 preceding salt, and especially from caesium chloride, which 
 usually crystallizes with it when the conditions are right for 
 its formation. After a great many trials three crops which 
 were satisfactory were obtained for analysis. The third anal- 
 ysis represents crystals which were picked out of the solution 
 one at a time and separately pressed between smooth filter- 
 papers. All the preparations were carefully examined under 
 the microscope and were evidently pure. 
 
 Found. Calculated for 
 
 , v 6CsCLCu 2 Cl r 2HjO. 
 
 Caesium 64.77 65.07 64.09 64.10 
 
 Copper 9.38 9.42 10.04 10.20 
 
 Chlorine 22.70 22.83 . . . 22.81 
 
 Water (difference) . 3.15 2.69 3.14 2.89 
 
 The analyses show that the salt has the formula Cs 8 CuCl 4 . 
 H 2 O. 
 
 The previously described cuprous double halogen salts, with 
 the new caesium salts for comparison, are given below : 
 
356 THE CAESIUM-CUPROUS CHLORIDES. 
 
 Csesium Salts. Previous Salts. 
 
 CsCl.Cu 2 Cl a 4NH 4 C1.3Cu,Cl, 
 3CsCl.Cu,Cl, 2NH 4 I.Cu 2 I 2 .H 2 
 6CsCl.Cu 2 Cl 2 .2H 2 4KCl.Cu 2 Cl 2 
 4NH 4 Cl.Cu 2 Cl 2 
 
 It is remarkable that there is no correspondence in type 
 between the caesium compounds and the others, and that such 
 a variety of types appears to exist. The formula 4NH 4 C1. 
 3Cu 2 Cl a may be considered somewhat doubtful on account of 
 its complex ratio, and because with one-fourth less ammonium 
 chloride it would correspond to the first caesium salt. 
 
 The salt Cs 3 CuCl6 is noticeable on account of its rather 
 complex formula and because it has the same ratio of caesium 
 to copper as the previously described cupric salt Cs 3 Cu 2 Cl 7 . 
 2H 2 O. Since the latter has a ratio that is unique among the 
 bivalent metal double halogen salts, a close structural relation 
 between the two compounds is suggested. 
 
 These caesium-cuprous chlorides show a decided lack of con- 
 formity with Remsen's law* concerning the composition of 
 double halides. Two out of three of them fail to correspond 
 to the law, while one of these, instead of not containing more 
 than one CsCl for one CuCl, actually contains three times as 
 much caesium chloride as Remsen's law allows. 
 
 My thanks are due to Mr. L. C. Dupee, who prepared and 
 analyzed one sample of the salt CsCu 2 Cl 3 . 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 September, 1893. 
 
 * Amer. Chem. Jour., xi, 296 ; xiv, 85. 
 
ON THE DOUBLE CHLOEIDES AND BROMIDES OF 
 CAESIUM, RUBIDIUM, POTASSIUM, AND AMMO- 
 NIUM WITH FERRIC IRON, WITH A DESCRIPTION 
 OF TWO FERRO-FERRIC DOUBLE BROMIDES. 
 
 BY P. T. WALDEN. 
 
 PREVIOUS investigation on the double ferric halides seems 
 to have been devoted exclusively to the chlorides, and the 
 metal caesium has not as yet been worked with in this 
 connection. 
 
 In view of these facts it appeared desirable to prepare, as 
 far as possible, a complete series of the double halogen salts 
 of the above-named metals. Only negative results were 
 obtained, however, when attempts were made to prepare 
 double iodides, so that the work was necessarily confined to 
 the chlorides and bromides. 
 
 The following compounds have been previously described : 
 
 Rb 8 FeCl,j K 2 Fe01 6 H 2 
 
 .... (NH4) 2 FeCl 6 H 2 
 
 .... Na 2 FeCl 6 H 2 
 
 The existence of the above potassium and ammonium salts 
 has been confirmed in the present investigation, but the com- 
 pound Rb 8 FeCl, described by Godenroy,t could not be made, 
 although a hydrous salt of the same type was prepared with 
 caesium. A most careful series of experiments using large 
 quantities of the constituent chlorides was made in the 
 attempt to prepare the rubidium salt just mentioned. It is 
 not believed to be possible that Godeffroy obtained this com- 
 pound, and his error was probably due to his neglecting the 
 
 * Amer. Jour. Sci., xlviii, October, 1894. 
 t Arch. Pharm. [3], ix, 343. 
 
358 DOUBLE CHLORIDES AND BROMIDES, ETC., 
 
 water of crystallization in the salt Rb 2 FeCl 6 H 2 . There is not 
 a great difference between the theoretical composition required 
 by a 3 : 1 anhydrous compound and the 2 : 1 salt with one 
 molecule of water, especially as far as the chlorine and iron 
 are concerned. This can be seen from the following com- 
 parison: 
 
 Calculated for Calculated for 
 
 RbsFeCle. Rb 2 FeCl 5 H 2 O. 
 
 Eubidium . . . 48.83 40.44 
 
 Iron 10.65 13.29 
 
 Chlorine .... 40.52 42.01 
 
 Water .... __ L _ L _ L _ 4.26 
 
 100.00 100.00 
 
 Since the hydrous 2 : 1 salt is easily prepared, it therefore 
 seems certain that this must have been the single salt de- 
 scribed by Godeffroy. 
 
 The following is a list of the salts obtained : 
 
 3 : 1 Type. 2 : 1 Type. 1 : 1 Type. 
 
 Cs 8 FeCl 6 .H 2 O Cs 2 FeCl 5 .H 2 CsFeCl 4 4H 2 
 
 Cs 2 FeBr 5 .H 2 O CsFeBr 4 
 
 Kb 2 FeCl 6 .H 2 O 
 
 Eb 2 FeBr 6 .H 2 
 
 K 2 FeCl 6 .H 2 0* 
 
 (NH 4 ) 2 FeCl 6 .H 2 0* 
 
 NH 4 FeBr 4 .2H 2 
 
 It will be noticed that the type 2 : 1 is the most frequently 
 recurring, being found in every case except with potassium 
 and ammonium bromides. The salts of this type, as might be 
 expected, are also the most stable and easily made, especially 
 with caesium chloride, where it is formed through a very wide 
 range of conditions, leaving only a narrow margin for the other 
 two members of the series. It is remarkable, in view of these 
 facts, that this type should not have been obtained with 
 
 t These two salts have been previously described by Fritzsche, J. prakt 
 Chem., xviii, 483. 
 
WITH FERRIC IRON. 359 
 
 ammonium bromide, while the 1 : 1 type, which is compara- 
 tively unstable in other cases, is made without difficulty. 
 
 This investigation furnishes another striking example of the 
 fact, already noticed several times in this laboratory, that 
 caesium halides form more complete series of double salts than 
 the halides of the other alkali metals. With caesium chloride 
 we get a complete series, while with the chlorides of the 
 other alkali metals only one type appears. In the bromides 
 no double ferric potassium salt could be isolated, whereas two 
 well-defined and comparatively stable compounds were ob- 
 tained with caesium. 
 
 Wells and Campbell * have called attention to the fact that, 
 in a number of cases, double halides show an increase in ease 
 of formation from the iodides to the chlorides. No better 
 illustration could be had of this truth than the series of salts 
 prepared in the present investigation, where the chlorides were 
 made in greater number and with more ease than the bromides, 
 while no iodides at all could be obtained. 
 
 Preparation. All these salts were made by mixing solu- 
 tions of the simple halides, evaporating and cooling to crystal- 
 lization. It was found necessary in all cases to use solutions 
 slightly acidified with the corresponding halogen acid, in order 
 to prevent the formation of basic salts. A record, as nearly 
 exact as possible, was kept of the relative quantities of the 
 constituents used, and this has been indicated under each salt. 
 The crystals were freed from the mother-liquor by pressing 
 between smooth filter papers, and in every case where it was 
 admissible they were further dried by exposure to the air of 
 the laboratory. Where the salt was at all deliquescent it was 
 at once removed to a tightly stoppered tube whose weight had 
 been previously determined and weighed without loss of time. 
 In this manner a quite unstable body could be analyzed and 
 satisfactory results obtained. The purity of all the simple 
 alkali halides was tested with the spectroscope before using. 
 The very pure rubidium chloride used for this work was fur- 
 nished to this laboratory for the encouragement of scientific 
 
 * Amer. Jour. Sci., HI, xlvi, 432. 
 
360 DOUBLE CHLORIDES AND BROMIDES, ETC., 
 
 investigation by the firm of E. Merck of Darmstadt, through 
 their agents, Messrs. Merck & Co. of New York, and our 
 thanks are due to them for their unsolicited generosity. 
 
 Method of Analysis. Iron was weighed as Fe 2 O 3 in all 
 cases, after having been separated from the alkali metal by 
 precipitation with ammonia. The filtrate from this precipita- 
 tion was evaporated to dryness, the alkali metal converted into 
 sulphate and weighed as such after ignition in a stream of air 
 containing ammonia. Ammonium was estimated by distilling 
 with a solution of potassium hydroxide, absorbing the NH 8 liber- 
 ated in hydrochloric acid and determining its amount by alka- 
 limetry. Water was determined by combustion behind sodium 
 carbonate and absorption in a washed calcium chloride tube. 
 With (NH 4 ) 2 FeCl 6 H 2 O the water was removed by subjecting 
 the salt to a temperature of 150 C. in an air bath for one hour. 
 
 The Double Chlorides. These salts are all red except 
 CsFeCl4jH 2 O, which is straw yellow. There is a distinct 
 gradation in the shades of the salts of the type 2 : 1 from 
 (NH 4 ) 2 FeCl6H 2 O, which is a deep ruby red, growing lighter 
 through the caesium, rubidium, and potassium compounds until 
 the last is very nearly the color of potassium dichromate. 
 
 3:1 Ccesium Ferric Chloride, Cs s FeCl 6 H 2 0. This salt is 
 the only one of the 3 : 1 type which was prepared in the 
 present investigation. It separated from a solution containing 
 50 g. of caesium chloride and from .5 g. up to 2.5 g. of ferric 
 chloride. The following analyses were made from separate 
 crops : 
 
 Found> Calculated for 
 
 Caesium . . . 
 Iron .... 
 
 A. 
 
 . 58.30 
 791 
 
 a 
 58.42 
 785 
 
 c. 
 
 Cs s FeClH 2 O. 
 
 58.17 
 17 
 
 Chlorine . . 
 Water . . . 
 
 . 30.87 
 . 2.74 
 
 30.82 
 2.72 
 
 30.98 
 2.64 
 
 31.01 
 2.65 
 
 99.82 99.81 100.00 
 
 In color it closely resembles sodium dichromate. It is well 
 crystallized in small prisms which are arranged in compact 
 clusters radiating from a centre. 
 
WITH FERRIC IRON. 
 
 361 
 
 2 : 1 Ccesium, Rubidium, Potassium, and Ammonium Fer- 
 ric Chlorides, Cs z FeOl,H z O, Eb 2 FeCl 5 H,0, K 2 FeCl 6 H 2 0, and 
 (NH<\FeCl & H 2 0. Ii solutions of the several alkali chlo- 
 rides containing 50 g. each be made, it is necessary to add 
 3 g. of ferric chloride to make the caesium salt of this type, 
 10 g. to make the rubidium salt, 15 g. to make the potassium 
 salt, and 70 g. to make the ammonium salt. The caesium, 
 rubidium, and potassium compounds can be recrystallized 
 unchanged, although with the last two there is a tendency to 
 separate simple alkaline chlorides at the same time. Several 
 separate crops were analyzed of each salt with the results 
 shown below: 
 
 Caesium . . 
 Iron . 
 
 
 rouna. 
 
 -- 
 
 Calculated for 
 
 . . 51.15 
 . . 11.05 
 
 B. 
 
 51.05 
 10.98 
 34.19 
 3.59 
 
 c * CsjFeCl 5 HjO. 
 51.40 
 
 10.82 
 34.02 34.30 
 3.48 
 
 Chlorine 
 Water . . 
 
 Rubidium . 
 Iron . 
 
 . . 34.36 
 . . 3.55 
 
 100.11 
 
 Foi 
 A. 
 
 . . 40.51 
 . . 13.28 
 
 99.81 
 
 and. 
 B. 
 
 40.69 
 13.33 
 41.92 
 4.20 
 
 100.00 
 
 Calculated for 
 Bb,FeCl 6 H 2 0. 
 
 40.44 
 
 13.29 
 42.01 
 4.26 
 
 Chlorine 
 Water . . 
 
 Potassium . 
 
 . . 41.91 
 . . 4.23 
 
 99.93 
 
 Foi 
 A. 
 
 . . 23.66 
 . . 16.86 
 
 100.14 
 
 and. 
 B. 
 
 23.54 
 
 16.99 
 53.35 
 5.96 
 
 100.00 
 
 Calculated for 
 K,FeCl 8 HjO. 
 
 23.73 
 
 16.98 
 53.84 
 5.45 
 
 Chlorine 
 Water . . 
 
 Ammonium 
 
 . . 53.56 
 . . 6.20 
 
 100.28 
 
 99.84 
 
 Pound. 
 
 100.00 
 
 Calculated for 
 
 A. 
 
 . . 12.39 
 . . 19.13 
 
 B. 
 
 12.36 
 18.95 
 61.07 
 * 
 
 c * (JNH 4 ) 2 Fe01 8 H s O. 
 
 12.00 12.52 
 19.48 
 61.22 61.74 
 6.26 
 
 Chlorine 
 Water . . 
 
 . . 61.21 
 . . 7.39 
 
 100.12 
 
 100.00 
 
362 DOUBLE CHLORIDES AND BROMIDES, ETC., 
 
 All these salts are well crystallized in short prisms. The 
 caesium and rubidium compounds are permanent in the air, but 
 the potassium and ammonium salts absorb moisture quite 
 
 rapidly. 
 
 1:1 Ccesium Ferric Chloride, CsFeCl^.\H z O. This was 
 made from a solution containing 50 g. of caesium chloride and 
 180 g. of ferric chloride. Below are the analyses of separate 
 crops : 
 
 Found - Calculated for 
 
 Caesium . 
 
 A. 
 
 . . . 38.39 
 . . . 17.03 
 
 B. 
 
 38.53 
 16.85 
 
 c. 
 
 CaFeCUiB 
 
 39.39 
 16.48 
 
 Chlorine 
 Water . 
 
 . . . 41.76 
 . . . 3.14 
 
 41.73 
 3.63 
 
 41.98 
 4.03 
 
 41.77 
 2.36 
 
 
 100.32 
 
 100.74 
 
 
 100.00 
 
 This salt absorbs moisture in the air so rapidly that con- 
 siderable difficulty was experienced in preparing samples for 
 analysis. It is regarded as containing half a molecule of water 
 on the evidence of the analytical results, although it is not 
 unreasonable to suppose that all the water found was absorbed, 
 especially as the bromide, CsFeBr 4 , is anhydrous. The crys- 
 tals were slender needles, which rapidly lost their yellow 
 color in the air, turning red. 
 
 The Double Bromides. These are all very dark green, 
 almost black and quite opaque. Like the chlorides, the 2 : 1 
 caesium salt is darker than the rubidium compound of the 
 same type. As no corresponding potassium or ammonium 
 salt could be made, the comparison can be carried no farther. 
 The caesium and ammonium 1 : 1 bromides are of nearly the 
 same color. None of the double bromides are capable of 
 recrystallization. 
 
 # : 1 Ccesium and Rubidium Ferric Bromides, Cs^FeBr^O 
 and Kb 2 FeBr B H 2 0. The first of these salts was made with 
 the quantities of the constituent bromides about equal, the 
 second with 50 g. of rubidium bromide to 60 g. of ferric 
 bromide. The following are the analyses: 
 
WITH FERRIC IRON. 
 
 363 
 
 Caesium . . 
 
 Found. 
 A. B. 
 
 . . . 35.76 35.60 
 
 Iron . 
 
 . . . 8.05 7.93 
 
 Bromine 
 Water . . 
 
 . . . 54.20 54.15 
 . . . 2.52 2.84 
 
 Eubidium . 
 Iron . 
 
 100.53 100.52 
 
 Found. 
 A. B. 
 
 ... 26.20 26.14 
 . . . 8.86 8.99 
 
 Bromine 
 Water . . 
 
 . . . 62.13 62.12 
 . . . 2.90 2.88 
 
 
 100.09 100.13 
 
 Calculated for 
 
 35.95 
 7.56 
 
 54.05 
 2.44 
 
 100.00 
 
 Calculated for 
 
 26.51 
 
 8.68 
 
 62.02 
 
 2.79 
 100.00 
 
 Both compounds were obtained in short doubly terminated 
 prisms. The caesium salt is comparatively stable, while the 
 rubidium salt decomposes rapidly in the air. 
 
 1:1 Ocesium and Ammonium Ferric Bromides, CsFeBr^ 
 and NHiFeBr 4 H 2 0. A solution of 50 g. of caesium bromide 
 and 100 g. of ferric bromide gave the first of these salts in 
 slender needles. The second could not be obtained until 
 250 g. of ferric bromide had been added to 50 g. of ammo- 
 nium bromide. Separate crops of each were analyzed with the 
 results given below. 
 
 Found. Calculated for 
 
 Csesium . 
 
 . . 26 02 
 
 
 Iron . . . . 
 
 . . 11.25 11 30 
 
 
 Bromine . . 
 
 . . 63.01 62.99 
 
 
 
 100.28 
 
 Found. 
 
 
 Ammonium . . 
 Iron . . , 
 
 A. B. 
 
 . 3.98 3.92 
 . 13 48 13 59 
 
 c. 
 3.83 
 
 Bromine . . . 
 Water . . . 
 
 . 74.85 74.71 
 . 7.69* 7.78* 
 
 
 
 
 
 
 100.00 100.00 
 
 
 Calculated for 
 
 4.19 
 13.02 
 74.42 
 
 8.37 
 100.00 
 
 * Water by difference. 
 
364 DOUBLE CHLORIDES AND BROMIDES, ETC., 
 
 As the ammonium salt is so deliquescent that no satisfactory 
 determination of water was possible, it was taken by difference, 
 and it is believed that the results warrant the acceptance of 
 the formula as written above. Great care was exercised in an 
 attempt to prepare a 2 : 1 ammonium bromide, but without 
 success. NH 4 FeBr 4 2HaO and simple ammonium bromide were 
 finally crystallized out together in the same solution. This is 
 regarded as good evidence that no salt of a type higher in 
 ammonium exists. 
 
 Ferro-ferric Salts: ElBr.FeBr^FeBr z .3H z O and KBr.FeBr* 
 %FeBr z 3H z O. While endeavoring to obtain a double ferric 
 bromide with potassium, a dark green body separated from 
 a solution containing an excess of bromine, which gave a 
 black hydroxide with ammonia. This was considered so 
 remarkable that an effort was made to prepare corresponding 
 salts with the other alkali halides and ammonium, under the 
 same conditions. This attempt resulted in the formation of 
 only one other compound of the same kind, that with rubidium. 
 The ferrous iron in those salts was determined by titration 
 in the presence of hydrochloric acid with a standard potassium 
 dichromate solution. It was found to be impossible to deter- 
 mine water satisfactorily on account of the extreme instability 
 of both salts. It was therefore taken by difference. The 
 rubidium salt resulted from the bringing together in solution 
 of 50 g. of rubidium bromide and 150 g. of ferric bromide, the 
 potassium salt from a solution of 50 g. of potassium bromide 
 and 250 g. of ferric bromide. Below are the analytical 
 results, A, B, and C being separate crops. 
 
 Found. Calculated for 
 
 r ~~^~ ~~ ^ RbBr.FeBr 2 .2FeBr.3H 2 0. 
 
 Eubidium . . . 7.25 8.32 
 
 Ferrous iron . . 5.17 5.16 5.53 5.45 
 
 Ferric iron . . 11.10 10.71 10.13 10.90 
 
 Bromine . . . 68.83 68.37 68.48 T0.07 
 
 Water . . . 7.65* 5.26 
 
 100.00 100.00 
 
 * Water by difference. 
 
WITH FERRIC IRON. 365 
 
 Found. 
 
 Calculated for 
 KBr.FeBr,.2FeBr 3 .3H,O. 
 
 Potassium 3.47 3.92 3.98 
 
 Ferrous iron .... 4.31 4.26 5.71 
 
 Ferric iron .... 12.36 11.70 11.42 
 
 Bromine 73.15 73.09 73.39 
 
 Water 6.71* . . . 5.50 
 
 100.00 100.00 
 
 These salts are dark green in color and quite opaque, like 
 the double ferric bromides described above. The crystalliza- 
 tion of the rubidium salt is apparently rhombohedral, that of 
 the potassium cubical. 
 
 In conclusion the author wishes to express his sincere thanks 
 to Prof. H. L. Wells, under whose direction the work has been 
 carried on, for his kindly aid and many valuable suggestions. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 July, 1894. 
 
 * Water by difference. 
 
ON THE CESIUM-COBALT AND CAESIUM-NICKEL 
 
 DOUBLE CHLORIDES, BROMIDES, 
 
 AND IODIDES. 
 
 BY G. F. CAMPBELL. 
 
 As a continuation of the work in this laboratory on double 
 halogen salts, the investigation of the above-mentioned com- 
 pounds has been taken up. The study has been made in a 
 systematic manner with the view of preparing as complete a 
 series as possible. The salts obtained belong to three types, 
 and are as follows: 
 
 3 : 1 Type. 2 : 1 Type. 1 : 1 Type. 
 
 Cs 8 CoCl 6 Cs 2 CoCl 4 CsCoCl 8 .2H 2 
 
 Cs 8 CoBr 5 Cs 2 CoBr 4 
 
 .... Cs 2 CoI 4 
 
 .... .... CsNiCls 
 
 .... .... CsNiBr 3 
 
 The results show that cobalt forms double salts with much 
 greater facility than nickel, for with the latter metal only the 
 chloride and bromide of a single type could be obtained. 
 
 The series illustrates the increase in ease of formation of 
 double salts from the iodides to the chlorides, which has been 
 previously observed, especially in the case of the csesium-mag- 
 nesium salts by Wells and Campbell.* No csesium-nickel 
 iodide could be prepared. 
 
 It should be noticed that the two salts of the 3 : 1 type are 
 exceptions to Remsen's law concerning this class of bodies. 
 
 The previously described double halogen salts of cobalt and 
 nickel, as far as I have been able to learn, correspond to two 
 types of the caesium salts, and are as follows : 
 
 * Amer. Jour. Sci., xlviii, November, 1894. t Ibid., xlvi, 432. 
 
DOUBLE CHLORIDES, BROMIDES, AND IODIDES. 367 
 
 2 : 1 Type. 
 
 (NH 4 ) 2 CoF 4 .2H 2 
 (NH 4 ) 2 NiF 4 .2H 2 
 
 1 : 1 Type. 
 
 NH 4 CoCl 8 .6H 2 
 
 NH 4 NiCl 8 .6H 2 
 
 KCoF 8 .H 2 O 
 
 KNiF 8 .H 2 
 
 NaCoF 8 .H 2 
 
 NaNiF 8 .H 2 
 
 The following table gives approximately the composition of 
 the solutions from which the caesium salts under consideration 
 were crystallized by concentration and cooling : 
 
 Cs : Co or Ni (Atoms.) 
 
 Cs 3 CoCl 5 . , . 
 
 From 12 
 
 1 
 
 Cs 2 CoCl 4 . 
 
 " 6 
 
 1 
 
 CsCoCl 3 .2H 2 . . 
 
 0.4 
 
 1 
 
 Cs 3 CoBr 5 . . . 
 
 " 2 
 
 1 
 
 Cs 2 CoBr 4 . . . 
 
 " 1 
 
 1 
 
 Cs 2 CoI 4 , . . . 
 
 1 
 
 4 
 
 CsNiClg . . . 
 
 " 12 
 
 1 
 
 CsNiBr 8 . . . . 
 
 " 2.5 
 
 1 
 
 bo 6:1 
 0.4 : 1 
 
 1 " syrupy solution of CoCl 2 
 "1:1 
 
 " syrupy solution of CoBr 2 
 1 : 16 
 
 " syrupy solution of NiCl 2 
 " syrupy solution of NiBr 2 
 
 More or less of the corresponding halogen acid was present 
 in each case, and an increase of this was apparently equivalent 
 in effect to the addition of the caesium halide. In the case of 
 the two nickel salts, a rather large amount of the acid was 
 desirable, for if it was not present, the salts appeared only 
 upon heating the concentrated solutions and dissolved when 
 they cooled. 
 
 The color of the chlorides containing cobalt is a magnificent 
 blue, the bromides and the iodides containing the same metal 
 are green, while the two nickel salts are yellow. The two 
 nickel salts form almost microscopic crystals. The two salts 
 of the 3 : 1 type were obtained in crystals having a diameter of 
 about 5 mm., apparently combinations of the cube and octahe- 
 dron. The salts of the 2 : 1 type form large plates or prisms, 
 the habit evidently depending upon the composition of the 
 solutions from which they crystallize. The salt CsCoCl 8 .2H 2 O 
 forms rather small plates. Besides the blue salt just men- 
 
368 CESIUM-COBALT AND CAESIUM-NICKEL 
 
 tioned, a red csesium-cobalt chloride of the 1 : 1 type was 
 obtained which lost water of crystallization so readily, with 
 change of color, that it could not be analyzed in its original 
 condition. 
 
 The compound Cs 2 CoI 4 is deliquescent, while the other salts, 
 here described, are stable. All the salts are whitened when 
 brought into contact with water or alcohol, evidently on 
 account of decomposition. 
 
 The following analyses were made : 
 
 Found . . 
 Calculated . 
 
 Cs a CoCl 5 . 
 
 Caesium. 
 
 62.79 
 62.82 
 
 Cobalt. 
 
 9.16 
 9.24 
 
 Chlorine. 
 
 27.83 
 
 27.74 
 
 Found . 
 Calculated 
 
 Caesium. 
 
 56.86 
 56.99 
 
 Cobalt. 
 
 12.53 
 12.58 
 
 Chlorine. 
 
 30.40 
 30.43 
 
 Found . 
 Calculated 
 
 Caesium. Cobalt. 
 
 38.64 17.67 
 39.80 17.56 
 
 Chlorine. Water. 
 
 32.07 10.94 
 31.87 10.77 
 
 Found A . 
 B 
 
 . . . 46.65 
 
 " C . 
 
 Calculated . 
 
 . . . 45.81 
 . . . 46.52 
 
 Cobalt. 
 
 6.88 
 7.44 
 7.08 
 6.84 
 
 Bromine. 
 
 46.33 
 46.97 
 46.52 
 46.64 
 
 Found A 
 " B 
 " C 
 
 Calculated 
 
 Caesium. 
 
 41.21 
 
 41.26 
 
 Cobalt. 
 
 9.49 
 9.20 
 9.25 
 9.10 
 
 Bromine. 
 
 49.43 
 49.60 
 49.32 
 49.64 
 
DOUBLE CHLORIDES, BROMIDES, AND IODIDES. 369 
 
 Found A 
 " B 
 C 
 
 Calculated 
 
 Found A 
 " B 
 Calculated 
 
 Found A 
 " B 
 
 Calculated 
 
 Caesium. 
 
 29.69 
 
 31.93 
 
 44.42 
 
 44.61 
 
 CsNiBr s . 
 
 Caesium. 
 
 29.93 
 30.60 
 30.81 
 
 Cobalt. 
 7.10 
 
 7.34 
 7.31 
 7.09 
 
 Nickel. 
 19.70 
 
 19.14 
 19.66 
 
 Nickel. 
 
 13.83 
 13.58 
 13.58 
 
 Iodine. 
 60.24 
 
 60.29 
 60.98 
 
 Chlorine. 
 
 35.78 
 35.57 
 35.73 
 
 Bromine. 
 
 55.84 
 55.49 
 55.61 
 
 The author takes pleasure in expressing his indebtedness to 
 Prof. H. L. Wells for valuable advice in connection with this 
 investigation. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 September, 1894. 
 
 24 
 
ON THE DOUBLE HALIDES OF CESIUM, 
 
 RUBIDIUM, SODIUM, AND LITHIUM 
 
 WITH THALLIUM.* 
 
 BY J. H. PRATT. 
 
 IN previous investigations upon the double halides of triva- 
 lent thallium with the alkali metals, the salts of only potassium 
 and ammonium seem to have been carefully studied. The 
 only caesium and rubidium salts that have been made are 
 Cs 8 TlC1.2H 2 O and Rb 8 TlCl 6 .2H 2 O described by Godfrey,* 
 but in the present investigation the compounds of this type 
 were found to have one instead of two molecules of water of 
 crystallization. 
 
 The present research has been carried out very carefully 
 and systematically in order to obtain as complete a series of 
 double salts in each case as possible. The salts that have 
 been made belong to four types, corresponding to those pre- 
 viously made with potassium and ammonium, and are as 
 follows : 
 
 3:1 2:1 3:2 1:1 
 
 Cs,TlCl..H,0 Cs 2 TlC] 6 Cs 3 Tl 2 Cl 9 
 
 Cs 2 TlCl 6 .H 2 
 
 Cs 3 Tl 2 Br 9 CsTlBr 4 
 
 CsTlI 4 
 
 Kb 8 TlCl 6 .H 2 Eb 2 T101 5 .H 2 
 
 Kb 8 TlBr 6 .H 2 KbTlBr 4 .H 2 
 
 KbTlI 4 .2H 2 
 
 Na 8 TlCl 6 .12H 2 
 
 Li 8 TlCl 6 .8H 2 
 
 For comparison, a list of the previously described double salts 
 with potassium and ammonium is also given. 
 
 * Amer. Jour. Sci., xlix, May, 1895. f Landenberg's Handworterbuch. 
 
DOUBLE HALIDES WITH THALLIUM. 371 
 
 3:1 2:1 3:2 1:1 
 
 K 8 T1C1 6 .2H 2 K 2 T1C1 6 .3H 2 K 8 T1 2 C1 9 .1^H 2 KTlBr 4 
 
 (NH 4 ) 8 T1C1 6 .2H 2 K 8 Tl 2 Br 9 .liH 2 KT1I 4 .H 2 
 
 (NH 4 ) 8 T1C1 6 (NH 4 )TlBr 4 .5H 2 
 
 (NH 4 )TlBr 4 .2H 2 
 
 (NH 4 )TlBr 4 
 
 (NH 4 )T1I 4 
 
 Several points of interest, already noticed in connection 
 with double salts prepared in this laboratory, are well illus- 
 trated by the series of new compounds to be described. With 
 caesium, a more complete series of salts was prepared than 
 with the other alkali metals ; and there is also an increase in 
 ease of formation and in number of salts, from the iodides to 
 the chlorides. The salts, formed from the alkali metal with 
 the lower atomic weight are generally more soluble in water, 
 form in larger crystals and with more water of crystallization 
 than those with higher atomic weight. 
 
 Preparation. The double salts were prepared in each case 
 by mixing solutions of the thallic halide with the alkali halide 
 in widely varying proportions, evaporating and cooling to 
 crystallization. With the bromides and iodides the conditions 
 for obtaining the double salts were improved by the presence 
 of a little free bromine and iodine. 
 
 The crystals, soon after forming, were removed from the 
 solutions, quickly pressed between filter papers to remove the 
 mother-liquor, and, with the exception of the sodium and 
 lithium salts, allowed to stand exposed to the air for some 
 time. The latter, on account of their instability, were placed 
 in tightly stoppered weighing-tubes as soon as they were free 
 from the mother-liquor. 
 
 Method of Analysis. In determining thallium, the salt was 
 dissolved in warm water and a slight excess of ammonium 
 sulphide added to precipitate the thallium as thallous sulphide. 
 This was filtered and washed with water containing a little 
 ammonium sulphide. The precipitate was then dissolved in 
 hot dilute nitric acid, the solution evaporated with sulphuric 
 acid in a platinum crucible, and then heated to constant weight 
 
372 DOUBLE HALIDES OF CAESIUM, RUBIDIUM, 
 
 within a porcelain crucible over a small flame. The nitrate 
 from the thallous sulphide precipitation was evaporated with 
 sulphuric acid, the ammonium salts driven off, and the residual 
 alkali sulphate ignited in a stream of air containing ammonia. 
 The halogens were determined as silver salts in separate por- 
 tions, with the precaution of adding sulphurous acid in the 
 case of the iodides to prevent loss of iodine in dissolving, and 
 it was found to be necessary in all cases to use a large excess 
 of nitric acid in order to obtain the silver halide in a pure con- 
 dition. Water was determined by igniting in a combustion 
 tube, behind a layer of dry sodium carbonate, in a stream of 
 dry air and collecting it in a weighed calcium chloride tube. 
 
 3:1 Ccesium and Rubidium Thallic Chlorides, Cs s TlCl 6 . 
 H 2 and Rb s TlCl 6 .H z O. The ceesium salt is obtained, as a 
 white precipitate, when 0.25 g. of thallic chloride is added to 
 a solution of 50 g. of caesium chloride. The precipitate dis- 
 solves somewhat slowly upon heating the solution, and crystal- 
 lizes out on cooling. The range of conditions is very narrow 
 as 3 g. of thallic chloride to 50 g. of caesium chloride give the 
 salt, Cs 3 TlCl 9 . The salt is soluble in hot water, but Cs 3 Tl 2 Cl 9 
 crystallizes from the solution. 
 
 The rubidium salt has a much wider range of formation. It 
 is obtained when 1.5 to 25 g. of thallic chloride are added to a 
 solution of 40 g. of rubidium chloride. It is very soluble in 
 cold water, but gives another salt, Rb 2 TlCl 6 .H 2 O upon crystal- 
 lization. Both salts are white, as are all the chlorides with 
 one exception. Two separate crops of each were analyzed with 
 the following results : Found . 
 
 . Calculated for 
 
 A. B. 
 
 Caesium . . . 48.44 48.05 48.33 47.84 
 
 Thallium . . . 24.21 24.45 24.37 24.46 
 
 Chlorine . . . 25.37 25.53 . . . 25.54 
 
 Water .... 2.74 1.97 . . . 2.16 
 
 Found. Calculated for 
 
 A. B. Rb 8 TlCl 6 .H 8 0. 
 
 Rubidium ...... 36.54 37.09 
 
 Thallium . . . 29.02 29.65 29.50 
 
 Chlorine. . . . 30.99 31.17 30.81 
 
 Water .... 2.51 1.72 2.60 
 
SODIUM, AND LITHIUM WITH THALLIUM. 373 
 
 The caesium salt was obtained in hair-like crystals, too small 
 for measurement. The rubidium salt crystallized in thin plates 
 having a rhombic outline. Under the microscope these showed 
 an extinction parallel to the diagonals and in convergent light 
 a bisectrix at one side of the field, with the plane of the optic 
 axes at right angles to the longer diagonal, indicating mono- 
 clinic symmetry. 
 
 2 : 1 Ccesium and Rubidium Thallic Chlorides, Cs^TlCl^ 
 Cs 2 TWl B .ff 2 0, and Eb z TWl 5 .ff 2 0. The anhydrous caesium 
 salt is formed when 5 to 8 g. of thallic chloride are added to a 
 somewhat concentrated solution of 100 g. of caesium chloride, 
 and the hydrous salt, when 8 to 15 g. of thallic chloride are 
 added to a more dilute solution of 100 g. of caesium chloride. 
 The rubidium salt was observed when 1.25 to 18 g. of rubidium 
 chloride were added to a rather concentrated solution of 30 g. 
 of thallic chloride. The two hydrous salts are white and the 
 anhydrous compound is pale green. The caesium salts are 
 readily soluble in hot water, but the salt Cs 8 Tl 2 Cl 9 crystallizes 
 from the solution. The rubidium salt recrystallizes unchanged 
 from water. The following analyses were made upon separate 
 crops : 
 
 Found. 
 
 Calculated for 
 
 A. B. C8 2 T1C1 6 . 
 
 Osmm 40.46 40.17 41.07 
 
 Thallium . . . 31.11 31.82 31.62 31.52 
 
 Chlorine . . . 27.19 27.30 27.20 27.41 
 
 Water 81 ... .81 
 
 The small amount of water found in the above analyses, 
 equivalent to about one-fourth of a molecule, was probably 
 held mechanically by the crystals. 
 
 Found. 
 ., v Calculated for 
 
 A. B. C. C8,T1C1 5 .H,0. 
 I. II. 
 
 Osium . . 40.03 39.84 40.30 39.85 39.97 
 
 Thallium . . 30.75 30.71 31.11 30.98 30.65 
 
 Chlorine . . 26.85 . . . 26.56 26.93 26.67 
 
 Water . . , . 2.88 2.37 2.71 
 
374 DOUBLE HALIDES OF CAESIUM, RUBIDIUM, 
 
 Found. 
 A. B. 
 
 29.09 28.97 
 
 Calculated for 
 Rb 2 TlCl 5 .H 2 O. 
 
 29.97 
 
 35.94 
 
 35.74 
 
 35.76 
 
 30.74 
 
 30.97 
 
 31.11 
 
 3.34 
 
 
 
 3.16 
 
 Rubidium 
 Thallium 
 Chlorine 
 Water 
 
 The crystals of Cs 2 TlCl 6 were in needles too small for 
 measurement. 
 
 The crystallization of 
 Cs 2 TlCl 5 .H 2 O and Rb 2 
 T1C1 6 .H 2 O is orthorhom- 
 bic. The salts are 
 similar in habit and are 
 developed as in Figs. 1 
 and 2. The forms ob- 
 served are as follows : 
 
 a, 100 
 m 110 
 
 e, 102 
 
 The crystals of the csesium salt were only about .4 to .6 mm. 
 in length, but the faces were smooth and gave good reflections 
 on the goniometer. The axial ratio is, 
 
 a : I : b = 0.6762 : 1 : 0.6954. 
 
 JA 4 Oil A OT1 
 m Am, 110 A 1TO 
 m A a, 110 A 100 
 
 a A e, 100 A 102 
 m A d, 110 A Oil 
 <*A e, Oil A 102 
 
 e A e, 102 A T02 
 
 Crystals of the rubidium salt were obtained from about 1.5 
 to 4 mm. in length. The axial ratio is, 
 
 a -.I-. c = 0.6792: 1:0.7002. 
 
 Measured. 
 
 Calculated 
 
 *70 
 
 
 ... 
 
 *68 
 
 22' 
 
 ... 
 
 34 
 
 3' 30" 
 
 34 11' 
 
 62 
 
 51' 
 
 62 44' 
 
 71 14'; 
 
 71 16' 
 
 71 12' 
 
 43 
 
 9' 
 
 43 16' 
 
 54 
 
 6' 
 
 54 32' 
 
SODIUM, AND LITHIUM WITH THALLIUM. 375 
 
 Measured. Calculated. 
 
 A OT1 *69 36' 
 
 m A m, 110 A 1TO *68 7' ... 
 
 m A a, 110 A 100 34 4'; 34 9'; 34 5' 34 4' 
 
 a A e, 100 A 102 62 52' 62 49' 
 
 m A d, 110 A Oil 71 26> ; 71 23' 71 21' 
 
 d* e, Oil A 102 43 19' 43 4 
 
 e A , 102 A T02 54 15' 54 22' 
 
 3 : 2 Ccesium Thallic Chloride, Cs s Tl 2 Cl 9 . ThQ conditions 
 under which this salt can be made are very wide ; .5 to 29 g. 
 of caesium chloride form a heavy white precipitate when added 
 to a solution of 40 g. of thallic chloride. This dissolves read- 
 ily in the solution upon heating and crystallizes in slender 
 hexagonal prisms terminated by the pyramid. When the ratio 
 of the csesium chloride to the thallic chloride is 30 g. to 50 g., 
 a salt is obtained which crystallizes in hexagonal plates. 
 Analyses of the plates do not agree very closely with theory, 
 but it is evident that they are the same as the prismatic salt 
 with another crystalline habit. The high percentage of 
 csesium and the corresponding low percentage of thallium is 
 probably due to the slight inclusions held by the crystals, 
 which could be seen with the microscope. This salt is white, 
 permanent in the air, and recrystallizes unchanged from water. 
 The analyses given below are of separate crops made under 
 very different conditions. 
 
 Caesium. Thallium. Chlorine. Water. 
 
 A .... 34.93 ... ... .65 
 
 B . . . . 35.09 35.64-35.51 28.09-27.99 
 
 C ....... ... 28.06; .95 
 
 D ....... 35.63 
 
 E . . . . 35.03 35.69 28.06 
 
 F (Plates) . 36.64 33.85 28.15 
 
 G (Plates) . 36.18 34.46 28.18 .61 
 
 The water found in these analyses was probably held 
 mechanically by the crystals. 
 
376 DOUBLE HALIDES OF CESIUM, RUBIDIUM, 
 
 The prismatic variety of this salt showed only the forms of 
 the prism, 1010, and pyramid, 1011. 
 
 Axis c = 0.82566 ; 0001 A 10T1 = 43 37' 50" 
 
 Measured. Calculated. 
 
 p^p, lOIlAOlTl *40 21' 
 
 m*p, 10TO A 10T1 46 21 J' ; 46 22' 46 22' 
 
 Sections parallel to the basal plane show in convergent 
 polarized light the normal uniaxial interference figure, with 
 weak negative double refraction. The crystals served very 
 well as 60 prisms for the determination of the indices of 
 refraction with the following results : 
 
 Red, Li. Yellow, Na. Green, Tl. 
 
 <o= 1.772 1.784 1.792 
 
 = 1.762 1.774 1.786 
 
 3:1 Rubidium Thallic Bromide, 1$. 275r 6 .5, 0. Tffis 
 salt was formed, when 1.5 to 24 g. of thallic bromide were 
 added to a very concentrated solution of 50 g. of rubidium 
 bromide. It crystallizes in beautiful golden yellow crystals, 
 which are very soluble in water, giving the 1 : 1 salt on re- 
 crystallizing. Careful efforts were made to obtain a 2 : 1 and 
 3 : 2 rubidium thallic bromide, but without success. Several 
 separate products, made under very different conditions, were 
 analyzed with the results which follow: 
 
 Bromine. Water. 
 
 49.29 2.49 
 49.66 
 
 49.42 
 50.28 
 
 50.49 
 
 50.08 1.88 
 
 The somewhat high percentage of rubidium and the low 
 percentage of thallium found in the first four analyses is prob- 
 
 A .... 
 
 Rubidium. 
 
 . 28.57 
 
 Thallium. 
 
 B . . . 
 
 
 2039 
 
 C . . 
 
 28.18 
 
 2059 
 
 D . . . . 
 
 2803 
 
 2016 
 
 E . . . . 
 
 2770 
 
 2033 
 
 F . . . 
 
 
 2064 
 
 G . . . . 
 
 . 26.56 
 
 21.17 
 
 Calculated for 
 Kb 8 TlBr 6 .H 2 
 
 1 26.76 
 
 21.28 
 
SODIUM, AND LITHIUM WITH THALLIUM. 377 
 
 ably due to the large excess of rubidium 3 
 
 bromide in the concentrated solutions from 
 which the crystals were obtained. As more 
 thallic bromide was added, better crystals were 
 obtained in more dilute solutions, which give 
 percentages agreeing very well with the cal- 
 culated. 
 
 The crystallization of this salt is tetragonal. 
 Doubly terminated crystals were obtained up 
 to a length of 6 mm. 
 
 The forms observed are : 
 
 a, 100 m, 110 ft 111 
 
 c, 001 e, 101 
 
 The habit is shown in Fig. 3. 
 
 Axis b = 0.80728 ; 001 A 101 = 38 54' 45" 
 
 Measured. Calculated. 
 
 e A e, T01 A 101 *77 
 
 a A e,100 A 101 51 6'; 51 2'; 51 3' 51 5^' 
 
 a A p, 100 A 111 57 52' ; 57 54' ; 57 53' 57 52' 
 
 e Aft 101 A 111 32 5' ; 32 12' 32 8' 
 
 c Aft 001 A 111 48 51' ; 48 55' 48 46' 
 
 m Aft 110 A 111 41 7' ; 41 4' 41 13' 
 
 The crystals show a weak negative double refraction. 
 
 3 : 2 Ccesium Thallic Bromide, Cs 8 Tl z Br 9 . This salt was 
 observed, as yellowish red crystals, when 1 to 15 g. of thallic 
 bromide were added to a solution of 50 g. of caesium bromide. 
 It was always obtained in small striated crystals, which were 
 not adapted for measurement. It is permanent in the air and 
 recrystallizes unchanged from water. Analyses of separate 
 products gave the following results : 
 
 Fo " nd - Calculated for 
 
 Caesium . 
 
 A. 
 
 B. 
 
 26.52 
 
 c. 
 26.14 
 
 D. 
 
 Thallium 
 Bromine 
 
 . . 27.36 
 . . 47.24 
 
 27.21 
 47.14 
 
 27.28 
 47.08 
 
 47.27 
 
 26.13 
 26.72 
 47.15 
 
378 DOUBLE HALIDES OF CJESIUM, RUBIDIUM, 
 
 1 : 1 Ccesium and Rubidium Thallic Bromides, CsTlBr 
 and RbTlBr^.HzO. These two salts are of nearly the same 
 color, pale yellow. The rubidium compound, which retains its 
 lustre and color much better than the other, recrystallizes 
 unchanged from water, while the caesium salt gives Cs 3 Tl 2 Br 9 , 
 when its solution is evaporated to crystallization. The caesium 
 salt was observed when 2 to 10 g. of caesium bromide were 
 added to 40 g. thallic bromide, and the rubidium salt when 3 to 
 24 g. of rubidium bromide were added to 40 g. thallic bromide. 
 Analyses of several different crops gave the following results : 
 
 Found - Calculated for 
 
 CsTlBr 4 . 
 
 20.25 
 31.05 
 48.70 
 
 Calculated for 
 RbTlBr 4 .H 2 O. 
 
 13.63 
 32.51 
 
 50.99 
 
 2.87 
 
 The crystallization of these two salts is isometric, the cube 
 being the only form observed. 
 
 1 : 1 Ccesium and Rubidium Thallic Iodides, CsTlI and 
 Rb TlIi.H z 0. Both of these salts were prepared from solu- 
 tions containing a large excess of thallic iodide and also from 
 solutions containing a large excess of the alkali iodide, so that 
 no other type of double iodides with these two metals could be 
 obtained. As the thallic iodide was very difficultly soluble in 
 water, alcoholic solutions were used where the thallic iodide 
 was in excess. The salts are ruby red, with a brilliant lustre, 
 which is slowly lost in the air. Both are decomposed by water. 
 The analytical results obtained from several different crops are 
 given below: 
 
 "* Calculated for 
 
 CsBsium 
 
 A. 
 
 . . 19.14 
 
 B. C. 
 
 D. 
 
 20.44 
 
 Thallium 
 Bromine 
 
 . . 32.36 31 
 47.76 
 
 .79 32.04 
 . . 48.39 
 
 Found. 
 
 48.88 
 
 Kubidium 
 Thallium . 
 Bromine . 
 Water . 
 
 A. 
 
 . . . 13.77 
 . . . 32.18 
 . . . 50.06 
 3.80 
 
 B. 
 
 13.41 
 
 * 
 
 c. 
 13.91 
 
 50^30 
 
 A B. 
 
 Caesium .... 16.57 16.38 . . . 15.74 
 
 Thallium .... 24.09 24.04 . . . 24.14 
 
 .... 59.48 59.67 60.12 
 
SODIUM, AND LITHIUM WITH THALLIUM. 379 
 
 Found. 
 
 Kubidium 
 Thallium 
 Iodine . 
 Water . 
 
 10.34 
 24.98 
 
 60.38-60.32 
 4.50 
 
 B. 
 
 9.78 
 25.23 
 60.79 
 
 Calculated for 
 RbTU 4 .2H,O. 
 
 10.26 
 
 24.47 
 
 60.94 
 
 4.32 
 
 These salts crystallize in the isometric system, the habit being 
 usually the cube truncated by the octahedron. 
 
 3 : 1 Sodium and Lithium Thallic Chlorides, Na s TlCl 6 .W 
 H z and Li z Tl Ol Q .8ff 2 0. Only one type of double salts could 
 be obtained with these metals, and it does not seem possible 
 that others exist, for the ground was covered very carefully 
 and systematically. On account of the extreme solubility of 
 these salts, especially that of the lithium compound, the 
 solutions had to be kept very concentrated, in a more or less 
 syrupy condition, which accounts for the high alkali metal and 
 low thallium found. These salts are transparent and colorless 
 when first taken from the mother-liquor, but, upon exposure 
 to the air, the sodium salt becomes opaque and the lithium 
 compound deliquesces. Analyses of different products gave 
 the following results : 
 
 Found. 
 
 Sodium 11.13 
 
 Thallium .... 27.79 
 
 Chlorine .... 31.23 
 Water 
 
 Lithium 
 Thallium 
 Chlorine . 
 Water 
 
 B. 
 
 10.48 
 28.39 
 30.45 
 29.75 
 
 Calculated for 
 
 Found. 
 
 1 
 
 A. 
 
 . 3.71 
 . 34.51 
 . 36.09 
 25.14* 
 
 B. 
 
 3.79 
 
 c. 
 3.73 
 
 D. 
 
 3.78 
 
 36.01 
 
 36.40 
 
 36.31 
 
 9.83 
 29.06 
 30.34 
 30.77 
 
 Calculated for 
 Li 8 TlCl.8H,0. 
 
 3.61 
 35.06 
 36.59 
 24.74 
 
 On account of the instability of the sodium and lithium 
 salts no crystallographic determinations were made. 
 
 Repeated attempts to prepare lithium and sodium thallic 
 
 By difference. 
 
380 DOUBLE HALIDES WITH THALLIUM. 
 
 bromides were entirely without success, hence no attempt was 
 made to prepare the iodides. 
 
 The author wishes to express his indebtedness to Prof. H. L. 
 Wells for valuable advice in connection with the chemical part 
 of this work, and to Prof. S. L. Penfield for suggestions con- 
 cerning the crystallography. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 December, 1894. 
 
ON THE DOUBLE SALTS OF CESIUM CHLORIDE 
 WITH CHROMIUM TRICHLORIDE AND WITH 
 URANYL CHLORIDE.* 
 
 BY H. L. WELLS AND B. B. BOLTWOOD. 
 
 NEUMANN! has made an extensive investigation of the 
 double salts formed with chromium trichloride and the chlo- 
 rides of several other metals, not, however, including caesium. 
 He obtained a violet double salt in each case with ammonium, 
 potassium, rubidium, beryllium, and magnesium, corresponding 
 to the general f ormula, 2M'C1. CrCl 8 .H 2 O, while with lithium, 
 sodium, calcium, strontium, barium, zinc, and cadmium he was 
 unable to prepare any double compounds. The double 
 fluorides, 2NH 4 F.CrF 8 .H 2 O, and 2KF.CrF 8 .H 2 O, which are 
 analogous to Neumann's salts, have been mentioned by Wag- 
 ner, J who also prepared the compounds 4NaF.2CrF 8 .H 2 O 
 and 3NH 4 F.CrF 3 . The existence of the latter salt has been 
 confirmed by Petersen. 
 
 Since Neumann had not prepared any caesium-chromium 
 chloride, and because, from the well-known comparative insolu- 
 bility of caesium double salts, it seemed possible that a greater 
 variety of compounds would be obtained with this than with 
 other metals, we have undertaken an investigation in this 
 direction. As the result of a systematic search, however, we 
 have added only a variation in water of crystallization to 
 Neumann's general formula. 
 
 Two salts have been obtained. One of these, 2CsCl.CrCl s . 
 H 2 O, is violet in color, corresponding exactly to Neumann's 
 compounds, while the other, 2CsCl.CrCl 8 . 4H 2 O, is green. 
 The violet salt was prepared by saturating warm aqueous 
 solutions containing various proportions of the two simple 
 chlorides with gaseous hydrochloric acid. The green salt was 
 
 * Amer. Jour. Sci., 1, 1895. t Liebig's Annalen, ccxliv, 329. 
 
 | Berichte, xix, 896. J. prakt. Chem., II, Ix, 52. 
 
382 DOUBLE SALTS OF CESIUM CHLORIDE 
 
 obtained from cold solutions by the use of hydrochloric acid, 
 and without its use by evaporation over sulphuric acid. 
 
 The salt 2CsCl.CrCl s .H 2 O forms aggregates of very minute 
 crystals of a magnificent red-violet color. It is stable in the 
 air and does not lose its water at 160. It is very slowly 
 soluble in cold water, forming a green solution from which 
 the green salt is deposited upon evaporation at ordinary temper- 
 atures. The four crops analyzed were prepared with amounts 
 of caesium chloride and chromic chloride varying from 15 g. of 
 the first and 50 g. of the second to 50 g. of the first and 10 g. 
 of the second. Gaseous hydrochloric acid caused a deposition 
 of the salt from warm solutions. The products, after careful 
 drying with paper and over sulphuric acid, gave the following 
 results upon analysis : 
 
 F< T d - Calculated for 
 
 2CsCl.CrCl s .H 2 0. 
 
 
 A. 
 
 B. 
 
 c. 
 
 D. 
 
 Caesium . . 
 
 50.31 
 
 49.72 
 
 49.64 
 
 . . . 
 
 Chromium 
 
 10.44 
 
 10.53 
 
 10.68 
 
 10.70 
 
 Chlorine . . 
 
 34.65 
 
 34.77 
 
 34.37 
 
 ... 
 
 Water . . . 
 
 4.11 
 
 5.12 
 
 
 
 
 99.51 
 
 10014 
 
 
 
 The salt 2CsCl.CrCl 3 .4H 2 O is deposited from cold concen- 
 trated solutions in the form of green, apparently monoclinic 
 crystals. It is somewhat deliquescent, very soluble in water, 
 and loses no water in the desiccator over sulphuric acid. At 
 110 it readily loses three molecules of water and is converted 
 into the violet salt. Three crops analyzed were prepared as 
 follows : Crop A, by evaporating a solution of 50 g. caesium 
 chloride and 25 g. of chromic chloride ; Crop B, by dissolving 
 the violet salt in water and evaporating over sulphuric acid ; 
 Crop C, by cooling a concentrated solution of 50 g. of each 
 chloride with the aid of ice and saturating it with hydrochloric 
 acid. The results were as follows : 
 
 Found - Calculated for 
 
 
 A. 
 
 B. 
 
 c. 
 
 2CsCl.CrCl s .< 
 
 Caesium . . . 
 
 . 46.40 
 
 46.13 
 
 46.73 
 
 46.86 
 
 Chromium . . 
 
 . 9.80 
 
 9.53 
 
 10.79 
 
 9.19 
 
 Chlorine . . . 
 
 . 31.30 
 
 31.14 
 
 . 
 
 31.27 
 
 Water .... 
 
 . ... 
 
 ... 
 
 
 
 12.68 
 
WITH CHROMIUM TRICHLORIDE, ETC. 383 
 
 A determination was also made of the water lost at 110 : 
 
 w^r,^ Calculated for 
 
 3H 2 in 2CsCl.CrCl 3 .4H,0. 
 
 Water 9.90 9.51 
 
 The variation in color of the two salts that have just been 
 described is interesting in connection with the violet and 
 green modifications of chromic salts in general, which have 
 furnished the ground for much investigation and discussion. 
 In the case under consideration the transformation from one 
 color to the other is accomplished by the addition or subtrac- 
 tion of water. It seems highly probable, however, that the 
 change in water is accompanied by a fundamental change in 
 the molecular structure, because the violet salt, containing 
 the smaller amount of water, is very much more slowly 
 soluble in water than the green salt, forming like the latter 
 a green solution. We have found that the whole of the chlo- 
 rine in the cold green solutions of these caesium salts is not 
 precipitated as silver chloride, thus showing that they agree 
 in this respect with other green solutions of chromic chloride. 
 
 It is a curious circumstance that the green chromic sul- 
 phate has been considered* to contain less water than the 
 violet modification, while with our caesium salts exactly the 
 reverse is true, the green salt containing the larger amount of 
 water. It is also remarkable that, while violet chromic solu- 
 tions are turned green by heat, our violet salt, nevertheless, 
 is produced hi hot solutions and the green salt in cold ones. 
 The theory advanced by Kriiger and maintained by Van Cleeff f 
 that the green color of chromic sulphate solutions is due to the 
 formation of a basic salt and free acid or an acid salt, seems 
 hardly applicable to the green caesium salts, since it crystal- 
 lizes from solutions saturated with hydrochloric acid in which 
 a basic salt would seem to be an impossibility. In view of 
 these apparently conflicting facts, it seems necessary to 
 draw the conclusion that the differences in color exhibited 
 by chromic compounds and their solutions are due to more 
 
 * Vide Van Cleeff, J. prakt. Chem., II, xxiii, 68. f Loc. cit. 
 
384 DOUBLE SALTS OF CESIUM CHLORIDE, ETC. 
 
 than one cause, probably to the formation of basic salts in 
 certain cases, and also, in other instances, to a change in water 
 of crystallization which is evidently accompanied by a molecu- 
 lar transformation. 
 
 Uranyl Chloride and Ccesium Chloride. A careful series 
 of experiments with caesium chloride and uranyl chloride 
 has resulted in the discovery of but a single salt. This 
 compound, 2CsCl.UO 2 Cl 2 , corresponds, except that it con- 
 tains no water, to the previously described salts, 2KC1. 
 U0 2 C1 2 .2H 2 0, 2KBr.UO 2 Br 2 .2H 2 O, 2NH 4 C1.UO 2 C1 2 .2H 2 O, and 
 2NH 4 Br.UO 2 Br 2 .2H 2 O, but some fluorides of other types 
 have been described. 
 
 The compound under consideration forms apparently ortho- 
 rhombic, yellow crystals which are usually small and blade-like 
 in shape. The products used for analysis were made under 
 the following conditions : Crop A, by making a concentrated 
 aqueous solution of 10 g. of caesium chloride and 50 g. of 
 uranyl chloride, then running in gaseous hydrochloric acid 
 until crystals began to form and cooling; Crop B, by the 
 same method as above, using 50 g. of caesium chloride and 
 10 g. of uranyl chloride; Crops C and D, by spontaneous 
 evaporation of solutions containing 50 g. of caesium chloride 
 and 15 g. of uranyl chloride ; and E, by the evaporation of 
 a solution of 15 g. of caesium chloride and 50 g. of uranyl 
 chloride. The results were as follows: 
 
 Fo " nd> Calculated for 
 
 AT^ B. C. ~~D~ ~1T 2C 8 C1.U0 2 C1 2 . 
 
 Cs . . 39.43 39.63 40.07 39.15 
 
 U0 2 . 40.37 41.14 40.96 41.85 43.39 39.95 
 
 Cl . . 20.63 21.17 20.85 20.84 20.59 20.90 
 
 The caesium chloride used in this investigation was from 
 a liberal supply of caesium and rubidium salts presented to 
 this laboratory, for the encouragement of scientific research, 
 by Herr E. Merck of Darmstadt, Germany, and we wish to 
 express our sincere thanks to him for his generosity. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 June, 1895. 
 
ON THE AMMONIUM-CUPROUS DOUBLE 
 HALOGEN SALTS.* 
 
 BY H. L. WELLS AND E. B. HURLBUBT. 
 
 THE existence of ammonium-cuprous double halides has long 
 been known, but since no complete investigation of these 
 compounds had been made, a careful study of them has 
 been undertaken. 
 
 Mitscherlich f prepared the potassium salt 4KCl.Cu 2 Cl 2 , 
 and mentioned the corresponding ammonium salt. This 
 salt, 4NH 4 Cl.Cu 2 01 2 , has been obtained in the present 
 investigation. 
 
 Deherain J described three double chlorides, 4NH 4 Cl.Cu 2 Cl2. 
 H 2 O, 2NH 4 Cl.Cu 2 Cl 2 , and NH.Cl.C^CL,. The first of these 
 salts, if the molecule of water is omitted, corresponds to the 
 compound mentioned by Mitscherlich which we have obtained, 
 and we are convinced that Deherain's formula for it is wrong. 
 The second salt, 2NH 4 Cl.Cu 2 Cl 2 , has not been obtained by us, 
 but since it corresponds in type to a bromide and an iodide 
 which are easily prepared, its existence seems possible. The 
 third salt of Deherain, NH 4 Cl.Cu 2 Cl 2 , probably does not ex- 
 ist, for we have failed to obtain it, as has Ritthausen also. 
 Ritthausen, while not being able to prepare NH 4 Cl.Cu 2 Cl 2 , 
 obtained the compound 4NH 4 C1.3Cu 2 Cl 2 , and we have con- 
 firmed this result. The compositions required for the two 
 formulae do not differ widely, so that it is probable that 
 Deherain analyzed the salt 4NH 4 C1.3Cu 2 Cl 2 and gave it an 
 incorrect formula. 
 
 As far as we know, no double bromides have been previously 
 described. Saglier || has described an ammonium-cuprous 
 
 * Amer. Jour. Sci., 1, 1895. t Ann. Chim. Phys., Ixxiii, 384. 
 
 J Compt. rend., Iv, 808. J. prakt. Chem., lix, 369. 
 
 II Compt. rend., civ, 1440. 
 
 25 
 
386 ON THE AMMONIUM-CUPROUS 
 
 iodide, to which the formula 2NHJ.Cu 2 I 2 .H 2 O is given. 
 The single double iodide which we have obtained corresponds 
 to Saglier's description and to his formula, except that we 
 have found it to be undoubtedly anhydrous. 
 
 In the present investigation a great number of experiments 
 have been made, with gradually varying proportions of the 
 constituent salts in each case, in order to obtain as many com- 
 pounds as possible. 
 
 The Chlorides, ^NH^Cl.Ou z Cl z and ^NH^Cl.3 Cu z Cl z . - 
 These compounds were prepared by making hot hydrochloric 
 acid solutions of mixtures of the simple salts, usually in the 
 presence of copper wire, and cooling to crystallization. The 
 first salt mentioned above is very readily oxidized by exposure 
 to air ; hence it has been found advisable in making it to use a 
 flask and to protect the solution from air by means of a stream 
 of carbonic acid. 
 
 The compound 4NH 4 Cl.Cu 2 Cl 2 requires the presence of a 
 comparatively large amount of ammonium chloride for its for- 
 mation, and crystallizes in colorless prisms which rapidly change 
 in color through brown to green upon exposure to the air. 
 Crystals 20 mm. in length and 5 mm. in thickness were observed. 
 
 The following analyses of two separate crops were made : 
 
 n> nnn j Calculated for 
 
 Found. 4NH.Cl.Cu.Cl* 
 
 Ammonium ..... 17.91 18.12 17.48 
 
 Copper 29.69 29.28 30.79 
 
 Chlorine 50.66 50.37 51.73 
 
 98.26 97.77 100.00 
 
 It was necessary to dry the samples for analysis very rapidly 
 on account of their instability, and some water was unavoidably 
 left in them, causing the low summations. The amount of 
 water corresponding to one molecule (Deherain's formula) is 
 4.19 per cent. 
 
 The other chloride, 4NH 4 C1.3Cu 2 Cl 2 , is produced when 
 the simple salts are mixed in the required proportion in hydro- 
 chloric acid solution, and also under considerable variations from 
 these proportions. It forms brilliant, colorless dodecahedra 
 
DOUBLE HALOGEN SALTS 387 
 
 which are moderately stable in the air at ordinary temperatures, 
 but gradually turn green on exposure. 
 
 The following analyses of three separate crops were made : 
 
 F( T d - Calculated for 
 
 Ammonium . 
 
 i. 
 . 9.39 
 
 n. 
 9.73 
 
 ILL 
 
 9.73 
 
 8.92 
 
 Copper . . 
 Chlorine . . 
 
 . 47.19 
 . 42.81 
 
 46.73 
 43.11 
 
 46.79 
 43.13 
 
 47.15 
 43.93 
 
 
 99.39 
 
 99.57 
 
 9065 
 
 100.00 
 
 The calculated amounts of ammonium, copper, and chlorine 
 for Deherain's formula, NH 4 Cl.Cu 2 Cl 2 , are 7.15, 50.50, and 42.35 
 respectively, and it does not seem possible that this formula 
 represents the true composition of the salt, because the samples 
 analyzed were well crystallized and evidently very pure. 
 
 The Bromides, 4NHBr. Cu 2 Br z and 2NH^Br.Cu z Br z .H z O. 
 - By the use of ammonium bromide, cuprous bromide, hydro- 
 bromic acid, and copper wire, these compounds were produced 
 similarly to the chlorides, but since these salts oxidize much 
 less readily than the chlorides, no protection by means of 
 carbon dioxide was necessary in any case. 
 
 The first salt, 4NH 4 Br.Cu 2 Br 2 , is formed in the presence 
 of an excess of ammonium bromide, and resembles the corre- 
 sponding chloride in form, occurring in long, colorless prisms 
 which turn green after long exposure to the air. Analyses of 
 two separate crops gave : 
 
 Found. Calculated for 
 
 I. II. 4NH 4 Br.Cu 3 Br,. 
 
 Ammonium 10.24 10.24 10.61 
 
 Copper 18.81 18.47 18.68 
 
 Bromine 70.93 70.60 70.71 
 
 99.98 99.31 100.00 
 
 The other bromide, 2NH 4 Br.Cu 2 Br 2 .H 2 O, is formed in 
 the presence of a relatively greater amount of cuprous bromide. 
 It forms brilliant, colorless rhombohedra, sometimes 15 mm. 
 long and 9 mm. wide, and it is more stable in the air than the 
 first bromide. Analyses of two separate crops gave : 
 
388 ON THE AMMONIUM-CUPROUS 
 
 Pound. Calculated for 
 
 I. II. 2NH 4 Br.Cu 2 Br,.H 2 0. 
 
 Ammonium 6.88 6.90 7.19 
 
 Copper 25.61 25.20 25.32 
 
 Bromine 63.76 64.08 63.90 
 
 Water (difference) . . . 3.75 3.82 3.59 
 
 The Iodide, 2NHJ. Ou 2 I 2 . Only one double iodide could 
 be obtained by the use of ammonium iodide and cuprous iodide 
 in widely varying proportions in hydriodic acid solutions. 
 This circumstance agrees with the observation made upon 
 several other series of double salts studied in this laboratory, 
 that the number of double salts possible decreases from the 
 chlorides to the iodides. Two separate crops gave the follow- 
 ing results upon analysis : 
 
 Found. Calculated for 
 
 I. II. 2NH 4 I.Cu 2 I 2 . 
 
 Ammonium 5.84 5.95 5.36 
 
 Copper 18.75 . . . 18.90 
 
 Iodine 75.07 75.55 75.74 
 
 99.66 100.00 
 
 Summary. The double salts obtained in the present inves- 
 tigation are as follows : 
 
 2 : 1 Type. 1 : 1 Type. 2 : 3 Type. 
 
 4NH 4 Cl.Cu 2 Cl 2 4NH 4 C1.3Cu 2 Cl 2 
 
 4NH 4 Br.Cu 2 Br 2 2NH 4 Br.Cu 2 Br 2 .H 2 
 
 2NH 4 I.CuA 
 
 The two bromides are apparently new compounds, while a 
 formula without water has been given to Saglier's iodide. The 
 compound, NH 4 Cl.Cu 2 Cl 2 , of Deherain probably does not 
 exist. 
 
 It was hoped that ammonium-cuprous salts of other types, 
 corresponding to the caesium-cuprous salts described by one of 
 us,* would be found, but such has not been the case, and there 
 
 * Amer. Jour. Sci., xlvii, 96. 
 
DOUBLE HALOGEN SALTS. 389 
 
 is no correspondence between the two series. The view 
 advanced in the article just mentioned, that the formula 
 4NH 4 C1.3Cu 2 Cl 2 might be considered somewhat doubtful on 
 account of its complexity and because its variation from the 
 1 : 2 type is slight, seems to have been unfounded. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 June, 1895. 
 
ON THE DOUBLE FLUORIDES OF CESIUM 
 AND ZIRCONIUM.* 
 
 BY H. L. WELLS AND H. W. FOOTE. 
 
 IN connection with his comprehensive work on zirconoflu- 
 orides, Marignacf has investigated the double fluorides of 
 zirconium with ammonium, sodium, and potassium, and since 
 the corresponding caesium salts have never been investigated, 
 we have undertaken a study of them. 
 
 The following table gives Marignac's ammonium and potas- 
 sium salts, together with those which we have prepared with 
 caesium : 
 
 3 : 1 Type. 2 : 1 Type. 1 : 1 Type. 2 : 3 Type. 
 
 3NH 4 F.ZrF 4 2NH 4 F.ZrF 4 
 
 3KF.ZrF 4 2KF.ZrF 4 KF.ZrF 4 .H 2 
 
 2CsF.ZrF 4 CsF.ZrF 4 .H 2 2CsF.3ZrF 4 .2H 2 
 
 The analogy between two types of caesium and potassium 
 salts is complete, while one type varies in each series. No 
 evidence has been found that caesium, in this case, forms a 
 greater variety of compounds than potassium. 
 
 The symmetrical arrangement of the vacancies in the table, 
 where no salts have been discovered, indicates that alkaline 
 fluorides of lower molecular weight combine with a relatively 
 smaller number of molecules of zirconium fluoride, while those 
 of higher molecular weight combine with a greater number of 
 such molecules. 
 
 The 2 : 1 type is the only one occurring in all three series. 
 This is the common and usually the only type of double halo- 
 gen salts formed by tetravalent elements ; hence its occurrence 
 in all cases was to be expected. The single sodium salt 
 
 * Amer. Jour. Sci., 1, 1896. f Ann. Chira. Phys., Ill, Ix, 257. 
 
DOUBLE FLUORIDES OF CAESIUM, ETC. 391 
 
 described by Marignac, 5NaF.2ZrF 4 , does not correspond to 
 any of the compounds in the above table, but it is to be 
 noticed that the composition corresponding to this formula 
 varies but little from that required for 2NaF.ZrF 4 . Although 
 Marignac's work on this salt was, as usual, very thorough and 
 careful, it seems possible that his products may have been the 
 2 : 1 salt containing a small amount of some impurity, possibly 
 a 3 : 1 compound. 
 
 Marignac described the salts Mn 2 ZrF f ,.6H 2 O, Cd 2 ZrF 8 . 
 6H 2 O, Zn 2 ZrF 8 .12H 2 O, and Cu 2 ZrF 8 .12H 2 O, all of which cor- 
 respond to a 4 : 1 type which has not been obtained with the 
 alkali metals. This type and those given in the preceding 
 table make five varieties of zirconofluorides, one of which has 
 been discovered in the present investigation. 
 
 The materials used for the preparation of the caesium salts 
 under consideration were carefully purified by ourselves. 
 Hydrofluoric acid was made from perfectly pure fluor-spar 
 and sulphuric acid, using a platinum still and redistilling the 
 product. Caesium carbonate, purified by the method described 
 by one of us,* was used in preparing the fluoride. Zircon was 
 used as the source of zirconium. The crude hydroxide was 
 conveniently obtained by fusing the finely pulverized mineral 
 with four parts of sodium carbonate, treating the resulting 
 mass with hydrochloric acid, evaporating with an excess of 
 sulphuric acid until the latter fumed, taking up with water, 
 filtering and precipitating with ammonia. For purifying the 
 zirconia, the method of Mitchell which has been advocated by 
 Baskerville f was found convenient. This consists in dissolv- 
 ing the zirconium hydroxide in hydrochloric acid, nearly 
 neutralizing with ammonia, adding a strong solution of sul- 
 phur dioxide and boiling. The precipitate, which, from the 
 results of Venable and Baskerville, \ appears to be a basic 
 zirconium sulphite, can readily be washed free from iron. 
 
 The double salts were prepared by mixing solutions of the 
 two fluorides in widely varying proportions, in the presence 
 
 * Amer. Jour. Sci., xlvi, 188. t Jour. Amer. Chem. Soc., xvi, 475. 
 
 J Ibid., xvii, 448. 
 
392 ON THE DOUBLE FLUORIDES OF 
 
 of more or less hydrofluoric acid, evaporating to the proper 
 point, and cooling. 
 
 When csesium fluoride is in excess, even with very small 
 amounts of zirconium fluoride, the salt 2CsF.ZrF 4 , is formed. 
 It crystallizes in rather large, simple hexagonal plates, showing 
 negative double refraction, and it can be recrystallized un- 
 changed from water. 
 
 When a larger proportion of zirconium fluoride is used, the 
 salt CsF.ZrF 4 .H 2 O is obtained. This forms monoclinic crys- 
 tals elongated in the direction of the b axis, and with faces 
 which are usually too rough for accurate measurement. This 
 salt also can be recrystallized unchanged from water. 
 
 With a large excess of zirconium fluoride extremely small, 
 difficultly soluble crystals of the salt 2CsF.3ZrF 4 .2H 2 O are 
 produced. The small crystals have a slight action upon polar- 
 ized light, but their form could not be made out. It does not 
 recrystallize from water in a pure condition, the product being 
 mixed with the 1 : 1 salt. 
 
 To determine csesium and zirconium, the fluorides were con- 
 verted into sulphates, then zirconium was separated from 
 csesium by the use of ammonia, and zirconium oxide and 
 csesium sulphate were finally weighed. In order to determine 
 fluorine a separate portion was dissolved in water, zirconium 
 hydroxide was precipitated with ammonia, sodium carbonate 
 was added in slight excess to the filtrate, and all the ammonia 
 was removed by evaporation. To the hot solution calcium 
 nitrate was added, and the resulting precipitate, after ignition, 
 was cautiously treated with dilute formic acid until, after 
 evaporation on the water-bath, a further addition of the acid 
 produced no effervescence. The calcium fluoride finally 
 remaining after a final evaporation was washed, ignited, and 
 weighed. The results of the fluorine determinations were 
 invariably somewhat low. 
 
 The substitution of formic acid for the acetic acid usually 
 used in removing calcium carbonate from the fluoride was sug- 
 gested by the greater volatility of the -first acid and the solu- 
 bility of its calcium salt. We have found the modification to 
 
CAESIUM AND ZIRCONIUM. 393 
 
 be an improvement as far as convenience is concerned, but we 
 are not yet prepared to say that it is more accurate than the 
 old method. 
 
 Water was determined by heating the substance in a boat 
 behind a layer of dry sodium carbonate in a combustion tube, 
 and collecting and weighing it in a calcium-chloride tube. 
 
 The following analyses of separate crops were made : 
 
 Calculated. 
 
 56.60 
 19.15 
 24.25 
 
 The small amount of water found in the analyses was evi- 
 dently mechanically included, for under the microscope bub- 
 bles of mother-liquor could be occasionally seen within the 
 crystals. 
 
 CsF.ZrF4.HtO. 
 
 ZCsF.ZrFt. 
 
 Found. 
 
 Caesium . . 
 Zirconium 
 Fluorine . . 
 Water . 
 
 A. 
 
 . . 56.41 
 . . 18.94 
 . . 22.73 
 1.63 
 
 B. 
 
 19.30 
 22.75 
 
 0.98 
 
 c. 
 55.51 
 19.16 
 
 0.97 
 
 
 JTOI 
 
 A. 
 
 ma. 
 B. 
 
 Calculated. 
 
 Caesium . . 
 
 
 
 38.44 
 
 39.58 
 
 Zirconium 
 
 . . 27.19 
 
 27.11 
 
 26.79 
 
 Fluorine . . 
 
 . . 27.24 
 
 27.52 
 
 28.27 
 
 Water . 
 
 6.27 
 
 5.20 
 
 5.36 
 
 
 JPOl 
 
 A. 
 
 ma. 
 B. 
 
 Calculated. 
 
 Caesium . 
 
 . . . 32.03 
 
 30.56 
 
 31.74 
 
 Zirconium 
 
 . . . 32.45 
 
 33.48 
 
 32.22 
 
 Fluorine . 
 
 . . . 31.09 
 
 30.43 
 
 31.74 
 
 Water . 
 
 4.40 
 
 3.96 
 
 4.30 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 July, 1895. 
 
ON CERTAIN DOUBLE HALOGEN SALTS OF 
 CAESIUM AND RUBIDIUM.* 
 
 BY H. L. WELLS AND H. W. FOOTE. 
 
 1. The Complicated Rubidium- Antimony Chloride. 
 
 REMSEN and SAUNDERS f have described a salt to which 
 they gave the formula 23RbC1.10SbCl 3 as the most probable 
 one. Wheeler, J working in this laboratory, confirmed Remsen 
 and Saunders' results, and discovered besides an analogous 
 bromide, to which the probable formula 23RbBr.lOSbBr 3 was 
 given. Remsen and Brigham prepared the salt 23RbCL 
 lOBiClg. Herty || has since described the two potassium salts 
 23KC1.10SbCl 8 and 23KBr.lOSbBr 8 .27H 2 O, and some mixtures 
 of these two salts. 
 
 In view of all this work, there can scarcely be a doubt as to 
 the existence of a type of salts with a somewhat complicated 
 ratio, but in view of the fact that this complicated ratio 23 : 10 
 is apparently an exception to the simplicity of composition of 
 all other carefully investigated double halogen salts, the sub- 
 ject seemed worthy of some further investigation. For the 
 purpose we have studied only the rubidium-antimony chloride 
 of Remsen and Saunders, as this salt is readily prepared and 
 is capable of repeated recrystallization from hydrochloric acid 
 solution. 
 
 The possibility suggested itself that the product might con- 
 sist of two simpler salts of similar or identical crystalline form, 
 which were capable of crystallizing together, and that previous 
 investigators had made use of conditions which resulted in 
 obtaining a constant mixture of two such salts. Although 
 
 * Amer. Jour. Sci., Ill, 1897. t Ibid., xiv, 155. 
 
 t Ibid., xlvi, 269. Amer. Chem. Jour., xiv, 174. 
 
 || Ibid., xvi, 490. 
 
CERTAIN DOUBLE HALOGEN SALTS. 395 
 
 this supposition had scarcely any probability in view of the 
 existence also of the rubidium-antimony bromide and of the two 
 potassium salts, we have put the question to test by repeatedly 
 recrystallizing the salt, using not only ordinary dilute hydro- 
 chloric acid for this purpose, but also more dilute and much 
 more concentrated acid and also an alcoholic hydrochloric acid 
 solution. As will be seen from the analyses, given beyond, no 
 variation in composition could be detected by the use of these 
 widely varying solvents for recrystallization, and it therefore 
 appears impossible that the salt can be a mixture. 
 
 As a starting-point, we used a solution in hydrochloric acid 
 containing the constituents RbCl and SbCl 3 in the exact molecu- 
 lar proportion 23 : 10. Product A was the first, B the third, 
 and C the fifth recrystallization from pure dilute hydrochloric 
 acid. The product D was obtained by adding concentrated 
 hydrochloric acid to a nearly saturated warm solution of the 
 salt in dilute hydrochloric acid. E was obtained from a very 
 strong hydrochloric acid solution formed by passing a rapid 
 current of hydrogen chloride gas into the solution as it cooled. 
 F was obtained by recrystallizing the salt from hydrochloric 
 acid, which was kept as dilute as it could be without producing 
 the basic double salt to be described beyond. G was a product 
 obtained by recrystallizing the salt from a mixture of equal 
 volumes of dilute hydrochloric acid and alcohol. 
 
 The two products obtained from concentrated hydrochloric 
 acid solution had a pale yellow color, while the others were all 
 white. The crystals were usually well-formed six-sided plates 
 which showed no definite optical properties. 
 
 The analyses of the various products are as follows : 
 
 Rubidium. Antimony. Chlorine. 
 
 A 39.23 23.85 37.01 
 
 B 39.23 23.84 36.99 
 
 C 23.91 
 
 D 39.25 23.98 
 
 E 39.31 23.89 
 
 F 39.03 23.86 
 
 G 39.11 23.90 
 
 Average .... 39.19 23.89 37.00 
 
396 CERTAIN DOUBLE HALOGEN SALTS 
 
 Method of Analysis. For the determination of antimony 
 and rubidium, a portion of about half a gram was dissolved 
 in water and enough hydrochloric acid to prevent antimony 
 oxychloride from precipitating. The solution was heated 
 to boiling and hydrogen sulphide passed in. The solution was 
 then cooled and the antimony sulphide filtered on a Gooch 
 crucible and washed with water and with alcohol. The crucible 
 was then slowly heated to 230 and cooled in an oven filled 
 with carbonic acid. The precipitate was weighed as Sb 2 S 3 . 
 The filtrate containing rubidium was evaporated with sulphu- 
 ric acid, and the residue ignited in a stream of air containing 
 ammonia and weighed as Rb 2 SO 4 . Chlorine was determined 
 by dissolving a separate portion in water acidified with tartaric 
 and nitric acids and precipitating with silver nitrate. This 
 was allowed to stand for some time, and the precipitate was 
 then collected on a Gooch crucible and weighed. .The methods 
 used are almost identical with those of Wheeler. 
 
 The accuracy of the antimony determination was checked in 
 the following manner. The salt Cs 3 Sb 2 Cl 9 was prepared from 
 very pure materials and carefully recrystallized, and antimony 
 determined by the above method. The per cent of antimony 
 is nearly the same as in the rubidium antimony salt under 
 consideration. The following results were obtained : 
 
 I. II. III. IV. 
 
 Per cent Sb found . . 25.37 25.42 25.43 25.44 
 " " calculated . 25.13 
 
 The atomic weights used in all the calculations were Rb, 85.43 ; 
 Sb, 120.43; Cl, 35.45; S, 32.07; Ag, 107.92; Cs, 132.89. 
 
 Since the method used for the determination x)f antimony 
 gives results which are slightly too high, we believe that a 
 deduction of the average error 0.25 per cent from the antimony 
 found in the analyses of the rubidium salt will 'give a result 
 which is nearer the truth. 
 
 Rb. Sb. Cl. 
 
 Average previously given . 39.19 23.89 37.00 
 
 Average with correction for Sb 39.19 23.64 37.00 
 
 Calculated for Rb 28 Sb 10 Cl 6S . ; 38.92 23.86 37.22 
 
 Calculated for Rb 7 Sb 8 Cl 16 . . 39.18 23.66 37.16 
 
OF CAESIUM AND RUBIDIUM. 397 
 
 It may be noticed that the results agree rather more satis- 
 factorily with the formula 7RbC1.3SbCl 8 than with the more 
 complicated one advanced by Remsen and Saunders. The 
 differences between these formulae are, however, so slight that 
 it is probably entirely impossible to decide between them by 
 means of chemical analysis, the ratio Rb : Sb being 230 : 100 
 in one case, and in the other 233 : 100. However, since it is 
 customary to use the simplest applicable formula for a chem- 
 ical compound, we propose the formula 7RbC1.3SbCl 8 for this 
 salt and corresponding formulas for other salts of this series. 
 Herty's hydrous salt, to which he gave the formula 23KBr. 
 10SbBr 3 .27H 2 O, agrees well with the formula 7KBr.3SbBr 8 
 8H 2 O. It must be admitted that the 7 : 3 ratio is an unusually 
 complicated one, but it is far simpler than 23 : 1 0, and is 
 scarcely a marked exception to the general simplicity of double 
 halogen salts. 
 
 2. A Rubidium-Antimony Oxy chloride, 2RbCl.SbCl 8 .SbOCl. 
 
 In attempting to recrystallize the salt 7RbC1.3SbCl 8 from 
 very dilute hydrochloric acid, just enough to prevent the forma- 
 tion of antimony oxychloride, this new salt was obtained in the 
 form of short colorless prisms possessing a rather high lustre. 
 It can be recrystallized from very dilute hydrochloric acid. 
 
 The following results were obtained from analyses of 
 separate crops : 
 
 Found. Calculated for 
 
 I. H. 2RbC1.8bCl 8 .8bOCL 
 
 Kb 26.54 26.68 26.68 
 
 Sb 37.58 37.36 37.61 
 
 Cl 32.75 32.80 33.21 
 
 O (by diff.) . . . 3.13 3.16 2.50 
 
 It is interesting to notice that Benedict * has described the 
 potassium salt 2KCl.SbCl 3 .SbOCl, which corresponds exactly 
 to this rubidium compound. 
 
 3. The Ccesium- Bismuth Chlorides and Iodides. 
 
 The double chlorides of bismuth with caesium have been 
 described by Remsen and Brigham. -f These authors did not 
 
 * Proc. Amer. Acad., xxix, 212. * Amer. Chem. Jour., xiv, 179. 
 
398 CERTAIN DOUBLE HALOGEN SALTS 
 
 state, however, how widely the conditions had been varied, and 
 we have repeated the work, varying the proportions of caesium 
 and bismuth as much as possible, and have found exactly the 
 same salts as described by them. These salts are, 
 
 3CsCl.BiCl s 
 3CsC1.2BiCl 3 
 
 SCsCLBiCls. This salt forms in colorless plates when 50 g. 
 of caesium chloride are mixed in hydrochloric acid solution 
 with from 1-25 g. of bismuth chloride. The analyses were 
 made on samples, dried but a short tune in the air, which 
 apparently contained a little mechanically included water. 
 The following results were obtained : 
 
 Found. Calculated for 
 
 I. II. 3CsCl.BiCl 3 . 
 
 Bi 24.80 24.47 25.36 
 
 Cs 47.94 . . . 48.66 
 
 Cl 25.98 
 
 3CsCl.2BiCh. When 50 g. of bismuth chloride are mixed 
 with from 1-80 g. of caesium chloride, the salt 3CsC1.2BiCl 3 
 crystallizes in light yellow needles, sometimes broadened and 
 looking like plates and again much shorter and thicker. 
 
 The following analyses were made : 
 
 Pound. Calculated for 
 
 I. II. 3CsC1.2BiCl 3 . 
 
 Bi 36.99 36.58 36.67 
 
 Cs 34.69 34.94 35.17 
 
 Cl 28.16 
 
 3C8L2Bil# Ccesium-Bismuth Iodide. 
 
 We could obtain only one double iodide of bismuth and 
 caesium, although the proportions of caesium and bismuth were 
 varied greatly. The salt formed as a crystalline precipitate, 
 difficultly soluble especially in an excess of caesium iodide, 
 when 1 g. of bismuth iodide was added to 50 g. of caesium 
 iodide and when 1 g. of caesium iodide was added to 50 g. of 
 bismuth iodide. With an excess of caesium, the color was a 
 bright red, while with an excess of bismuth the color was more 
 of a reddish brown. 
 
OF CESIUM AND RUBIDIUM. 399 
 
 Methods of Analysis. The methods here given were used 
 in both the double chlorides and iodide of bismuth. 
 
 Halogens were determined as the silver salts being precipi- 
 tated from a solution acidified with tartaric and nitric acids 
 and, after standing, filtered and weighed on a Gooch crucible. 
 As Remsen and Brigham had mentioned a difficulty in deter- 
 mining bismuth, we made a few determinations of it in Bi 2 O 8 , 
 which was made by precipitating BiONO 8 with water from a 
 nitric acid solution of Bi(NO 8 ) 8 , and heating the precipitate to 
 constant weight in a platinum dish. The method finally 
 adopted was to dissolve the substance in water slightly acidi- 
 fied with hydrochloric acid and precipitate Bi 2 S 8 from the cold 
 solution with hydrogen sulphide. The precipitate was filtered 
 and immediately dissolved in nitric acid and digested for some 
 time on the water bath until completely decomposed. The 
 sulphur was filtered off and the filtrate, diluted to about 300- 
 400 c. c., was heated and ammonium carbonate added in slight 
 excess. It was placed on the water bath for an hour or two, 
 until the liquid had become nearly clear and the excess of 
 ammonium carbonate had been driven off, and it was then fil- 
 tered on a Gooch crucible and ignited strongly over a Bunsen 
 burner and weighed as Bi 2 O 3 . 
 
 Two determinations on Bi 2 O 8 gave the following results : 
 
 I Amt. Bi 2 8 taken = 0.1979 g. Amt. Bi 2 8 found = 0.1974 g. 
 II = 0.3604 g. " " = 0.3617 g. 
 
 The filtrate from the bismuth precipitation was evaporated 
 with sulphuric acid and ignited in a stream of air containing 
 ammonia. The residue was weighed as Cs 2 SO 4 . 
 
 The results obtained from the analysis of the double iodide 
 were as follows : 
 
 Found. Calculated for 
 
 I. II. 
 
 Bi ..... 21.34 21.15 21.25 
 
 Cs ..... 20.75 20.31 20.38 
 
 I ........ 58.02 58.37 
 
 SHEFFIELD CHEMICAL LABORATOBY, 
 January, 1897. 
 
ON THE DOUBLE FLUORIDES OF ZIRCONIUM 
 WITH LITHIUM, SODIUM, AND THALLIUM.* 
 
 BY H. L. WELLS AND H. W. FOOTE. 
 
 IN a previous article f we have described the caesium-zir- 
 conium fluorides, and upon comparing these with the corre- 
 sponding ammonium and potassium salts, which had been 
 previously described by Marignac, J it was observed that the 
 types of salts formed varied with the molecular weights of the 
 alkaline fluorides in an interesting manner. The fluorides of 
 smaller molecular weight gave types with a larger relative 
 number of these molecules, while the fluorides of higher molec- 
 ular weights combined with more zirconium fluoride than the 
 others. This relation is made clear from the following table, 
 which was given in the previous article referred to : 
 
 2:3Type. 
 
 3:lType. 
 
 3NH 4 F.ZrF 4 
 3KF.ZrF 4 
 
 2 : 1 Type. 
 
 2NH 4 F.ZrF 4 
 2KF.ZrF 4 
 2CsF.ZrF A 
 
 1 : 1 Type. 
 
 KF.ZrF 4 
 CsF.ZrF, 
 
 2CsF.3ZrF 4 .2H 2 
 
 The present investigation was undertaken with the view, 
 in the first place, of testing the apparent rule with lithium 
 fluoride, which has a lower molecular weight than the fluorides 
 previously experimented upon. Our expectations were real- 
 ized by the preparation of the salt 4LiF.ZrF 4 .|H 2 O. The salt 
 2LiF.ZrF 4 was also obtained, but, in spite of a careful search, 
 no intermediate 3 : 1 salt could be discovered. The following 
 table, giving the lithium, potassium, and caesium salts, shows a 
 perfectly symmetrical gradation in types according to the 
 
 * Amer. Jour. Sci., Ill, 1897. t Ibid., IV. i, 18. 
 
 J Ann. Chim. Phys., ix, 257. 
 
DOUBLE FLUORIDES OF ZIRCONIUM. 
 
 401 
 
 atomic weights of the alkali metals, except that the intermedi- 
 ate lithium salt is missing. 
 
 Type. 
 
 4: 1 
 3 : 1 
 
 Lithium Salts. 
 
 4LiF.ZrF 4 .H 2 
 
 2:1 
 1 : 1 
 
 2LiF.ZrF 4 
 
 2:3 
 
 
 Potassium Salts. 
 (Marignac.) 
 
 3KF.ZrF 4 
 2KF.ZrF 4 
 
 Caesium Salts. 
 
 2CsF.ZrF 4 
 
 KF.ZrF 4 .H 2 CsF.ZrF 4 .H 2 
 
 2CsF.3ZrF 4 .2H 2 
 
 Marignac's two ammonium salts, 3 : 1 and 2 : 1, also enter 
 the series symmetrically. 
 
 We have investigated also the thallous-zirconium fluorides, 
 since the high atomic weight of thallium led us to expect that 
 it would possibly yield a series of salts symmetrical with those 
 of the alkali metals with a still higher ratio of zirconium than 
 was the case with caesium. Such was not the case, however. 
 The salts discovered were : 
 
 3TlF.ZrF 4 , 5TlF.3ZrF 4 .H 2 0, TlF.ZrF 4 , and TlF.ZrF 4 .H 2 0. 
 
 Two of these three types of thallous salts correspond to 
 types of alkali-metal salts, while one type, the 5 : 3, is a new 
 one, but the series is not symmetrical with the others accord- 
 ing to the atomic weights. 
 
 Since Marignac had described but one sodium-zirconium 
 fluoride, 5NaF.2ZrF 4 , and since this differs from all other 
 alkaline zirconium fluorides, we have undertaken a new inves- 
 tigation of the sodium salts. As a result, we have fully con- 
 firmed Marignac's results as to the 5 : 2 salt, which is the one 
 most readily obtained, and we have succeeded in preparing a 
 new salt, 2NaF.ZrF 4 , which corresponds to the most usual 
 type of double halogen salts of tetravalent elements. It is 
 evident, however, that the sodium salts, like those of thallium, 
 do not form a symmetrical series with the others. 
 
 The following table gives a list of the sodium and thallium 
 salts, and shows the positions, "X," of the other compounds 
 prepared by Marignac and ourselves. 
 
 26 
 
402 DOUBLE FLUORIDES OF ZIRCONIUM 
 
 Lithium Ammonium godium Potassium Caesium Thallium 
 
 Type. Salts. Salts. galt8 . , ^alts. Salts. Salts. 
 
 4:1 X 
 
 3 . 1 .. X X .. 3TlF.ZrF 4 
 
 2 1 X X 2NaF.ZrF 4 X X 
 
 5.3 5TlF.3ZrF 4 H 2 
 
 <TlF.ZrF 4 
 
 \ TlF.ZrF 4 .H 2 
 
 2:3 X 
 
 While our investigation has shown that the rule for the 
 variation of the types with the atomic weights applies only 
 partially to the zirconium double fluorides, we have shown at 
 least that the variety of types is remarkable, and it is also 
 noticeable that the ratios are nearly the simplest that can exist 
 in such number between the extreme limits 4 : 1 and 2 : 3. 
 
 Preparation. Thallium fluoride was prepared by dissolv- 
 ing the metal in sulphuric acid, adding an excess of baryta 
 water, filtering and passing carbonic acid into the hot solution. 
 The filtrate from this precipitation was evaporated and treated 
 with hydrofluoric acid in excess. The salts were prepared by 
 mixing the acid solutions of the fluorides in varying propor- 
 tions, evaporating and cooling to crystallization. The salts 
 were then removed and pressed between filter papers till dry. 
 In all cases they were stable in the air. 
 
 Method of Analysis. Zirconium and the alkalies were 
 determined by evaporating the salt with sulphuric acid to 
 drive off hydrofluoric acid, precipitating zirconium hydroxide 
 with ammonia and weighing ZrO 2 . The filtrate was evapo- 
 rated to dryness and the alkali determined as sulphate, either 
 by igniting with ammonium carbonate or heating in a current 
 of air containing ammonia. When thallium was present, the 
 fluoride was dissolved in water, a little sulphurous acid added 
 to make sure that the thallium was all in the univalent condi- 
 tion, and the zirconium precipitated with ammonia. The pre- 
 cipitate needed to be very thoroughly washed. The filtrate 
 was evaporated nearly or quite to dryness to remove free 
 
WITH LITHIUM, SODIUM, AND THALLIUM. 403 
 
 ammonia, diluted to a volume of about 100 c. c., heated to 
 boiling and potassium iodide added in excess to precipitate 
 thallium iodide. This was collected on a Gooch crucible, 
 washed with eighty per cent alcohol, dried at 100 C. and 
 weighed. Fluorine was determined by the ordinary calcium 
 fluoride method after precipitating zirconium with ammonia 
 and removing ammonium salts by evaporation with sodium 
 carbonate. Water was determined by heating the salt in a 
 combustion tube behind a layer of dry sodium carbonate and 
 collecting the water in a calcium chloride tube. 
 
 Salts of Lithium. 
 
 2LiF.ZrJF 4 . This salt forms when from 0.7 g. to 2 g. of 
 lithium fluoride are added to 20 g. of zirconium fluoride. The 
 crystals are hexagonal, showing prism and pyramid and rarely 
 a basal plane. In appearance, they are very much like crystals 
 of quartz from Herkimer County, New York, but they are 
 very small. On recrystallizing, the 4 : 1 salt was formed. 
 
 Separate crops were analyzed with the following results : 
 
 Li 
 Zr 
 
 F 
 
 Found. 
 
 I n. 
 6.03 6.39 
 
 Calculated for 
 Li 2 ZrF e . 
 
 6.42 
 
 41.81 41.64 
 
 41.28 
 
 51.62 
 
 52.30 
 
 . This was the most unsatisfactory salt 
 obtained, though it seems undoubtedly to establish the 4 : 1 
 type. As lithium fluoride is very insoluble, only a compara- 
 tively small amount could be dissolved in zirconium fluoride, 
 and apparently we could not go far enough toward the lithium 
 end to get the salt in pure condition. It formed in a crust 
 ordinarily, and the crystals were very small. Under the micro- 
 scope, no mixture with another salt could be found in the 
 crops analyzed. Once, however, it was obtained mixed with 
 the 2 : 1 salt, as seen under the microscope, showing there 
 could probably be no intermediate salt. Various conditions 
 were tried, and crops were obtained from both hot and cold 
 
404 DOUBLE FLUORIDES OF ZIRCONIUM 
 
 solutions. It forms when 5 to 7 g. of lithium fluoride are 
 mixed with 20 g. of zirconium fluoride. On recrystallizing, 
 lithium fluoride is precipitated. 
 
 Following are the results of the analyses: 
 
 I 
 
 II 
 
 III 
 
 IV 
 
 V 
 
 Salts of Sodium. 
 
 2NaF.ZrF 4 . This salt crystallizes in very minute crystals 
 of hexagonal outline, coming down in a crust when from one 
 to two parts of sodium fluoride are added to fourteen parts of 
 zirconium fluoride. The salt does not recrystallize. The fol- 
 lowing results were obtained from separate crops. The water 
 was probably mechanically included. 
 
 
 Li. 
 
 . . . 9.54 
 
 Zr. 
 
 33.14 
 
 H 3 0. 
 
 4.83 
 
 F. 
 
 
 
 
 4.93 
 
 
 [ 
 
 . . . 9.79 
 
 33.30 
 
 4.35 
 
 53.16 
 
 
 
 33.23 
 
 
 
 
 
 33.02 
 
 
 
 Iculated for Li,ZrF 8 . 
 
 H 2 . . 9.93 
 
 31.91 
 
 4.26 
 
 53.90 
 
 Found. Calculated for 
 
 I. H. 
 
 Na . . . 18.66 18.41 18.40 
 
 Zr . . . 34.78 36.21 36.00 
 
 H 2 . . . 1.96 0.50 
 
 F* . . . 44.60 44.88 45.60 
 
 . Marignac has previously described this salt, 
 which comes down under wide conditions in very good crys- 
 tals and recrystallizes easily. Prof. L. V. Pirsson has kindly 
 examined the crystals and made the following report : 
 
 " The crystals show good sharp forms, but are very small. 
 They appear distinctly orthorhombic in habit, consisting in the 
 main of rather stout prisms, made up of two prismatic planes, 
 m and m', and terminated by a rather steep brachydome. In 
 another habit, which is rarer, the front pinacoid, a, is broadly 
 developed, while the prisms are very small ; this type also 
 shows at times a pyramid, p. The plane of the optic axes lies 
 
 * By difference. 
 
WITH LITHIUM, SODIUM, AND THALLIUM. 405 
 
 in the base and a = c, b = %,c = b. The optic angle is large, 
 and it could not be told whether a or b was the acute bisectrix. 
 The double refraction is very low. The crystals in their form 
 strongly recall the figures of chrysolite (olivine) shown in the 
 mineralogies." 
 
 The analyses gave the following results from different crops : 
 
 Found. Calculated for 
 
 I. II. 
 
 Na ..... 21.15 21.09 21.23 
 
 Zr ..... 33.63 33.55 33.22 
 
 F* ..... 45.22 45.36 45.57 
 
 Salts of Thallium. 
 
 TlF.ZrF^O and TlF.ZrF^. These salts crystallize in 
 somewhat concentrated solutions when one part of thallium 
 fluoride is mixed with three or four parts of zirconium fluoride. 
 The analyses invariably show an excess of zirconium fluoride. 
 The hydrous salt crystallizes in needles, if the solution be 
 cooled before precipitation occurs. If the solution is evapo- 
 rated until crystals begin to form and then cooled, the anhy- 
 drous salt deposits in minute square plates. The salt gives 
 the 5 : 3 type on recrystallizing. The following results were 
 obtained : 
 
 Found. Calculated for 
 
 I. II. TlZrF 5 .H 2 0. 
 
 50.05 
 
 22.15 
 23.37 
 4.43 
 
 Calculated for 
 TlZrF s . 
 
 Tl ..... 50.16 49.91 52.37 
 
 Zr ..... 23.86 24.08 23.17 
 
 F ........ 24.32 24.46 
 
 5TlF.3ZrFi.HtO. This salt crystallizes in needles when 
 from one to three and one-half parts of thallium fluoride are 
 added to one part of zirconium fluoride. When about four 
 
 * By difference. 
 
 Tl 
 
 .... 48.43 
 
 47.91 
 
 Zr . 
 
 .... 22.93 
 
 23.16 
 
 F . 
 
 
 23.17 
 
 H 2 
 
 . 3.89 
 
 4.80 
 
406 DOUBLE FLUORIDES OF ZIRCONIUM. 
 
 parts of thallium fluoride are added, the same salt crystallizes 
 in a different habit, forming prisms of hexagonal outline which 
 under the microscope are seen to be twinned, resembling in 
 this respect the hexagonal-shaped crystals of aragonite. On 
 recrystallizing, both habits give the needle-shaped crystals. 
 
 The following analyses were made of the two kinds of crys- 
 tals. A rather large number of determinations was made on 
 account of the existence of two different forms. 
 
 TL Zr. H 2 0, P. 
 
 I ............ 61.58 16.88 ...... 
 
 II ............ 62.05 16.84 ...... 
 
 III ........... 61.37 17.14 1.40 . . . 
 
 IV ........... 61.58 16.88 1.17 19.31 
 
 V ............ 61.74 17.04 1.42 . . . 
 
 VI ................. 1.31 . . . 
 
 VII ........... 62.91 16.42 ...... 
 
 Calculated for Tl 6 Zr 8 F 17 .H 2 . . 62.47 16.58 1.11 19.84 
 
 ^ Crystals of this salt form in brilliant octa- 
 hedra when one part of zirconium fluoride is added to from 
 four to twenty parts of thallium fluoride. It is easily recrys- 
 tallized. 
 
 The following analyses were made : 
 
 Found. Calculated for 
 
 I. II. Tl s ZrF 7 . 
 
 Tl ..... 72.82 73.20 73.24 
 
 Zr ..... 10.91 10.38 10.80 
 
 F ..... 15.65 . . . 15.96 
 
 SHEFFIELD CHEMICAL LABORATORY, 
 January, 1897. 
 
ON THE CAESIUM ANTIMONIOUS FLUORIDES 
 
 AND SOME OTHER DOUBLE HALIDES 
 
 OF ANTIMONY. 
 
 BY H. L. WELLS AND F. J. METZGER. 
 
 WE have made a thorough study of the double salts formed 
 by caesium fluoride and antimonious fluoride with the result 
 that five compounds have been prepared. This is an un- 
 usual number for a series of double salts, and it gives a good 
 illustration of the facility with which caesium forms such 
 compounds. 
 
 The salts to be described have the following formulas : 
 
 Type. 
 
 1:3 CsF.3SbF 8 
 
 1:2 CsF.2SbF 8 
 
 4:7 4CsF.7SbF, 
 
 1:1 CsF.SbF 8 
 
 2:1 2CsF.SbF 8 
 
 The previously described antimonious double fluorides, all 
 of which were prepared by Fliickiger,* are as follows : 
 
 Type. 
 
 1:1... KF.SbFs .... .... 
 
 2:1 ... 2KF.SbF 8 2LiF.SbF 8 2NH 4 F.SbF 8 
 3:1... 3NaF.SbF 8 .... .... 
 
 The following chlorides, bromides, and iodides have been 
 described : 
 
 * Pogg. Ann., Ixxxvii, 245 (1852). 
 
408 CAESIUM ANTIMONIOUS FLUOEIDES AND 
 
 Type. 
 
 1 : 2 BbC1.2SbCl 3 * 
 
 3:4 3NH 4 I.4SbI 8 .9H 2 Ot 
 
 1 : 1 NH 4 Cl.SbCl 8 ; * NH 4 LSbI 8 .2H 2 ; KLSbI 8 H 2 ; 
 
 BbCLSbCl, II 
 
 ( 3KI.2SbI,.3H 2 ; t 3NaI.2SbI,.12H,O ; t 
 3:2 \ 3NH 4 L2SbI 8 .3H|0 1 3RbC1.2SbCl 8 ; * 
 
 (3RbBr.2SbBr 8 ;* 3RbL2SbI 8 ; * 3CsC1.2SbCl 8 1[ 
 
 2 : 1 2NH 4 Cl.SbCl, ; t 2NH 4 Cl.SbCl 8 .H 2 ; IT 2KClSbCl 8 ; ft 
 
 2KCl.SbCl 8 2H 2 
 
 1 : 3 7EbC1.3SbCl 3 ; || 7KbBr.3SbCl 3 j * 7KC1.3SbCl, ; || || 
 7KBr.3SbBr 8 .8H 2 || || 
 
 3 : 1 3NH 4 Cl.SbCl 8 .liH,0; ** (3KCl.SbCl,) ; ** 3NaCl.SbCl 8 * 
 
 4 : 1 4NH 4 I.SbI 8 .3H 2 Ot 
 
 Upon comparing the caesium double fluorides with the salts 
 already known, it is to be noticed that two types of the 
 former, 1 : 3 and 4 : 7, do not occur among the latter, and that 
 the 3:4, 3:2, 7:3, 3:1, and 4 : 1 types were not found 
 among the csesium antmionious fluorides. The absence of a 
 3 : 2 fluoride is remarkable, since the salt 3CsC1.2SbCl 3 is 
 very sparingly soluble, and because this is a very prominent 
 type among the chlorides, bromides, and iodides. It is evident 
 that a close relation does not exist between the caesium anti- 
 monious fluorides and the other antimonious double halides, 
 and that the types of the former could not have been pre- 
 dicted from a consideration of the latter. In range, the 
 
 * Wheeler, Amer. Jour. Sci., (Ill), xlvi, 269. 
 t Schaffer, Pogg. Ann., cix, 611. 
 t Deherain, Compt. rend., lii, 734. 
 Nickles, Ibid., li, 1097. 
 
 || Remsen and Saunders, Amer. Chem. Jour., xlv, 152. 
 If Setterberg, Of versigt K. Vetensk-akad. Forhandl., 1882, 23 ; Eemsen and 
 Saunders, loc. cit. 
 
 ** Poggiale, Compt. rend., xx, 1180. It is probable according to Herty, 
 Amer. Chem. Jour., xv, 81, that 3KCl.SbCl 3 does not exist. 
 
 tt Jacquelaine ; Poggiale, loc. cit. ; Benedict, Proc. Amer. Acad., xxix, 212. 
 tt Benedict, loc. cit. 
 
 This type was described as 23 : 10 ; see Wells and Foote, Amer. Jour. 
 Sci., iii, 461. 
 
 II || Herty, loc. cit. 
 
OTHER DOUBLE HALIDES OF ANTIMONY. 409 
 
 caesium double fluorides extend farther at the antimony end 
 than the others, while they do not extend as far at the alkali- 
 metal end of the series of types. 
 
 When all the types of antimonious double halides are con- 
 sidered, they are remarkable for their large number, ten. 
 This number is probably greater than is the case with any 
 other negative element. The types, 1 : 3, 1 : 2, 4 : 7, 3 : 4, 
 1:1, 3:2, 2:1, 7:3, 3:1, and 4:1, with two or three 
 exceptions, are the simplest that can exist in such number 
 between the two extremes, and arithmetically they extend 
 almost as far in one direction as the other. 
 
 Method of Preparation. Solutions of caesium fluoride and 
 antimonious fluoride were prepared by treating caesium car- 
 bonate and antimonious oxide, each with an excess of pure 
 hydrofluoric acid. To the antimonious solution the caesium 
 salt was gradually added in small quantities, and after each 
 addition the liquid was evaporated and cooled until crystalli- 
 zation took place. If a homogeneous product was obtained 
 a portion was removed for analysis, and the process was con- 
 tinued until finally the liquid contained a very large excess of 
 caesium fluoride. In every case the products were carefully 
 inspected to make sure that they were not mixtures, and at 
 least two crops of a salt were always prepared under some- 
 what different conditions, and were shown by analysis to be 
 identical in composition before they were accepted as true 
 compounds. 
 
 Method of Analysis. The crystals were carefully dried by 
 pressing between filter papers, and the portions to be analyzed 
 were preserved in glass weighing-tubes which were coated 
 within with a very thin layer of paraffine. For the deter- 
 mination of antimony and caesium a portion was heated in 
 a platinum crucible with concentrated sulphuric acid until 
 all the hydrofluoric acid was removed, the residue was dis- 
 solved in hydrochloric acid, antimony was precipitated as 
 sulphide, collected on a Gooch crucible and weighed after 
 drying in a small oven containing carbon dioxide. Caesium 
 was weighed as normal sulphate. Fluorine was determined 
 
410 CJESIUM ANTIMONIOUS FLUORIDES AND 
 
 by converting it into silicon fluoride, collecting the latter 
 in water, and titrating with sodium hydroxide, according to 
 a modification of Offermann's method. The results of these 
 fluorine determinations were invariably somewhat too low, as 
 we found by testing the method with pure potassium silicon 
 fluoride; hence the chief value of these determinations con- 
 sist in showing that the salts under investigation were not 
 oxy-compounds. 
 
 1 : 2 Ccesium Antimonious Fluoride, CsF.%SlF z . This salt 
 was obtained in the form of beautiful transparent needles by 
 adding 2 or 3 g. of caesium fluoride to a solution of about 50 g. 
 of antimonious fluoride in somewhat dilute hydrofluoric acid 
 solution, heating to boiling and cooling. Two separate crops 
 gave the following results upon analysis : 
 
 Calculated for 
 CsSb 2 F 7 . 
 
 Csesium 26.28 26.44 
 
 Antimony .... 47.43 47.36 47.42 
 
 Fluorine .... 26.28 25.23 
 
 1 : 3 Ccesium Antimonious Fluoride, CsF.3SbFs. This 
 salt crystallizes in the form of stout, transparent prisms. 
 It was obtained by evaporating the mother-liquors from the 
 preceding compound and cooling, or by the use of somewhat 
 less caesium fluoride in proportion to the antimonious fluoride 
 in a more concentrated solution. Two crops gave the follow- 
 ing results : 
 
 Calculated for Foun(L 
 
 CsSb 3 P 10 . r[ ^ 
 
 Csesium 19.47 17.81 17.60 
 
 Antimony .... 52.71 52.95 53.89 
 
 Fluorine .... 27.82 27.01 26.84 
 
 The rather wide variation of the results from the calculated 
 quantities is probably due to the fact that the crystals were 
 taken from a very concentrated antimonious fluoride solution 
 and were consequently not quite pure, even after careful dry- 
 ing on paper. 
 
OTHER DOUBLE HALIDES OF ANTIMONY. 411 
 
 4 7 Ccesium Antimonious Fluoride, 4C 8 F>7iSbF a . This 
 salt crystallizes in transparent plates, and is formed in the 
 presence of a little larger proportion of caesium fluoride than 
 the preceding compounds. It was obtained, for instance, upon 
 adding about 4 g. of caesium fluoride to a mother-liquor from 
 the last salt and crystallizing by cooling. Two crops gave the 
 following results: 
 
 Calculated for Found. 
 
 Caesium 28.80 28.99 
 
 Antimony .... 45.47 46.02 46.03 
 
 Fluorine .... 25.73 24.58 24.61 
 
 We cannot say that we are absolutely sure about the 
 formula of this apparently complicated double salt. It can- 
 not be a 1 : 2 compound, for not only is it entirely distinct 
 in appearance from CsF.2SbF 8 , but coming as it does from a 
 strong antimony solution, the results would naturally come 
 too high rather than too low for antimony. The results vary 
 too widely from a 2 : 3 ratio to make that probable, but they 
 approach somewhat more closely the 3 : 5 ratio. The follow- 
 ing calculations will show that we have selected the most 
 probable formula: 
 
 Calculated for Calculated for Calculated for 
 
 CB2Sb s F u . CsjSbgF^. CsSb,F 7 . 
 
 Cesium 31.86 29.75 26.28 
 
 Antimony .... 43.11 44.75 47.43 
 
 Fluorine .... 25.03 25.50 26.28 
 
 1 : 1 Caesium Antimonious Fluoride, CsF.SbF^. In the 
 presence of still greater proportions of caesium fluoride this 
 salt is produced by cooling the properly concentrated solution. 
 It forms square prisms, the ends of which are not usually 
 modified by any planes. Three crops gave the following 
 analyses : 
 
 Calculated for *^ 
 
 Cs8bF 4 . 
 
 Caesium . . . 40.43 41~44 41.19 
 
 Antimony . . 36.47 35.85 35.66 35.52 
 
 Fluorine. 23.10 22.30 
 
412 CAESIUM ANTIMONIOUS FLUORIDES AND 
 
 2 : 1 Caesium Antimonious Fluoride, %CsF.SbF s . This 
 salt is formed under a wide range of conditions when csesium 
 fluoride is present in large excess hi comparison with the anti- 
 monious fluorine. It crystallizes in apparently rhombic prisms, 
 which are often somewhat flattened. Four crops, made under 
 very different conditions, gave the following results : 
 
 Calculated for Foun<L 
 
 Cesium 
 
 5530 
 
 i. 
 
 5481 
 
 n. 
 
 in. 
 
 IV. 
 
 Antimony 
 Fluorine . 
 
 . 24.95 
 . 19.75 
 
 24.72 
 19.42 
 
 24.59 
 
 24.92 
 
 24.64 
 
 By the use of very concentrated caesium fluoride solutions 
 with comparatively small amounts of antimonious fluoride, 
 no evidence was obtained of the existence of any double 
 salts containing more csesium fluoride than the one just 
 described. 
 
 Ccesium Antimonious Iodide, 3CsI.SbI B . It appears 
 that no compound of csesium iodide with antimonious 
 iodide has been described. The sparingly soluble chloride, 
 3CsC1.2SbCl 8 is well known, and this was the only double 
 chloride that Remsen and Saunders * were able to prepare, 
 although four rubidium antimonious chlorides are known. 
 It is evident that the slight solubility of csesium antimoni- 
 ous chloride makes it impossible to prepare concentrated 
 solutions of the component chlorides, and consequently pre- 
 vents the formation of salts of other types. We have found 
 that an iodide which corresponds in composition to the chloride 
 can be readily prepared. It is sparingly soluble in hydriodic 
 acid solutions, and it exists in two distinct forms, one of 
 which is brick-red and apparently octahedral in form, while 
 the other is yellow and occurs in thin hexagonal plates. The 
 octahedral salt was prepared by mixing antimonious iodide and 
 csesium iodide in rather strong hydriodic acid solutions, while 
 the yellow hexagonal salt was made in much less strongly acid 
 solutions, particularly upon diluting them with water, boil- 
 
 * Loc. cit 
 
OTHER DOUBLE HALIDES OF ANTIMONY. 413 
 
 ing, and cooling. Two crops of each form were analyzed as 
 follows : 
 
 Found. 
 
 CasSbJa. Red Salt. 
 
 Yellow Salt. 
 
 
 
 i. 
 
 n. 
 
 I. 
 
 n. 
 
 Caesium . 
 
 . 22.39 
 
 23.46 
 
 . . . 
 
 22.15 
 
 ... 
 
 Antimony 
 
 . 13.47 
 
 13.91 
 
 13.19 
 
 14.36 
 
 14.52 
 
 Iodine . . 
 
 . 64.14 
 
 62.98 
 
 . . . 
 
 63.03 
 
 . . . 
 
 This was the only double iodide that we were able to obtain. 
 
 There is little doubt that a corresponding bromide exists, 
 for we observed, while engaged in work with another object in 
 view, that a yellow precipitate is produced when the bromides 
 of caesium and trivalent antimony are brought together in 
 solution. Having overlooked the fact that thejcompound had 
 not been described, we neglected to analyze the product. 
 
 Ccesium Antimonic Halides. So little is known concerning 
 the double halides of quinquivalent elements that it seemed 
 desirable to study the caesium antimonic compounds. Setter- 
 berg * has described a single double chloride, CsCl.Sb.Cl 5 , and 
 we have confirmed his result, but by using widely varying 
 conditions we have been unable to prepare any other compound. 
 Setterberg's salt crystallizes in long, colorless, transparent 
 needles. A crop of it gave the following results upon analysis : 
 
 Calculated for 
 
 Osium ..... 28.54 29.14 
 
 Antimony .... 25.75 26.43 
 
 Chlorine .... 45.70 43.94 
 
 We have extended our investigation to antimonic fluoride 
 and caesium fluoride, but the results were disappointing from the 
 the fact that we were able to prepare but one double salt, while 
 Marignacf has described two potassium antimonic fluorides. 
 Either caesium in this case unexpectedly fails to show as 
 great a tendency to form double salts as does potassium, or 
 else we have failed to find the proper conditions for produc- 
 ing them. 
 
 * Loc. cit. t Liebig's Ann., cxlv, 237. 
 
414 CESIUM ANTIMONIOUS FLUORIDES. 
 
 The salt obtained by us apparently contains hydroxyl, 
 although prepared in strong hydrofluoric acid solutions, and 
 has the formula CsF.SbF 4 OH. It crystallizes on cooling 
 warm, rather concentrated solutions in the form of bundles of 
 transparent needles. Two crops gave the following analyses : 
 
 Calculated for Fou . D(L 
 
 CsSbF 5 OH. T~ ~~n ' 
 
 Caesium 36.44 37.77 
 
 Antimony .... 32.87 31.82 31.72 
 
 Fluorine .... 26.03 25.54 26.18 
 
 Hydroxyl .... 4.66 (4.87) 
 
 LD SCIENTIFIC SCHOOL, 
 April, 1901. 
 
ON THE DOUBLE CHLORIDES OF (LESIUM 
 AND THORIUM. 
 
 BY H. L. WELLS AND J. M. WILLIS. 
 
 NEAKLY all of the known double halogen salts of quadri- 
 valent metals belong to a single type, of which 2KCl.PtCl 4 
 and 2KF.SiF 4 are examples. It has been shown, however, 
 by Marignac* and by Wells and Foote f that the double flu- 
 orides of zirconium exist in a variety of types. Therefore, 
 since thorium is somewhat closely related to zirconium, we 
 have undertaken an investigation of some thorium double 
 halides, and have selected the caesium salts as being the most 
 promising. 
 
 Upon attempting to prepare caesium thorium fluorides we 
 found that thorium fluoride is practically insoluble even in 
 concentrated solutions of caesium fluoride containing hydro- 
 fluoric acid. There is no doubt that the two fluorides combine 
 under these circumstances, but, since we obtained only finely 
 divided precipitates as products and there was no certainty as 
 to their purity, further work on the fluorides was abandoned. 
 Chydenius $ has previously described two potassium thorium 
 fluorides, 2KF.ThF 4 .4H 2 O and KF.ThF 4 .JH 2 O, but on account 
 of the insolubility of thorium fluoride and of these double salts 
 it is probable that there may be some doubt in regard to the 
 correctness of these formulas. 
 
 We have prepared two caesium thorium chlorides, to which 
 we assign the formulas 3CsCl.ThCl 4 .12H 2 O and 2CsCl.ThCl 4 . 
 11H 2 O. The amount of water of crystallization in these com- 
 pounds is somewhat uncertain, since they form very small 
 hygroscopic crystals, and it is difficult to dry them by pressing 
 
 * Amer. Chim. Phys., Ill, lx, 257. 
 t Amer. Jour. Sci., IV, i, 18; iii, 466. 
 t Pogg. Ann., cxix, 43. 
 
416 THE DOUBLE CHLORIDES 
 
 on paper. The search for double chlorides was made system- 
 atically by starting with a solution of about 65 g. of thorium 
 chloride in hydrochloric acid, adding 2 to 4 g. of csesium 
 chloride at a time, and evaporating and cooling after each 
 addition, until finally, after dividing the solution and using a 
 part of it, a very large excess of csesium chloride was present. 
 
 In analyzing the salts chlorine was determined as silver 
 chloride, sometimes in separate portions, in other cases in the 
 nitrates from which thorium hydroxide had been precipitated ; 
 thorium was weighed as oxide after precipitation with ammonia, 
 and the csesium in the filtrates was converted into normal 
 sulphate and weighed as such; water was determined by 
 difference. 
 
 3 : 1 Ccesium Thorium Chloride, 3 Os 01. Th Oli.WH* 0. This 
 salt was produced from solutions containing about 12 g. of 
 thorium chloride and from 30 to 110 g. of csesium chloride. 
 It forms colorless crystals of feathery structure upon cooling 
 very concentrated solutions. Three different crops made 
 under somewhat varied conditions gave the following results 
 upon analyses : 
 
 Caesium . . 
 Thorium 
 Chlorine 
 Water . . 
 
 2:1 Ocesium Thorium Chloride, 20sOl.ThCl^llH 2 0. - 
 This salt was obtained in colorless crystals, somewhat resem- 
 bling the previous salt, but not nearly as feathery in appearance. 
 It was formed in concentrated solutions containing about 65 g. 
 of thorium chloride and from 30 to 100 g. of csesium chloride. 
 The following analyses were made of different crops : 
 
 Calculated for 
 Cs 3 ThCl 7 .12H 2 O. 
 
 . 36.45 
 
 
 f ounu. 
 
 
 r i. 
 36.21 
 
 II. 
 
 36.14 
 
 in. 
 
 . 21.20 
 
 20.70 
 
 21.68 
 
 21.05 
 
 . 22.61 
 
 23.09 
 
 . . . 
 
 23.37 
 
 . 19.74 
 
 (20.00) 
 
 . . . 
 
 . . . 
 
 Caesium . 
 
 Calculated for 
 Cs 2 ThCl 6 .llH 2 0. 
 
 . 29.26 
 
 
 JTOU 
 
 na. 
 
 
 i. 
 
 29.84 
 
 II. 
 
 29.10 
 
 III. 
 
 28.92 
 
 IV. 
 
 Thorium . 
 
 . 25.52 
 
 25.42 
 
 25.41 
 
 25.70 
 
 25.22 
 
 Chlorine . 
 
 . 23.43 
 
 23.40 
 
 24.75 
 
 23.35 
 
 23.55 
 
 Water . 
 
 . 21.78 
 
 (21.34) 
 
 (20.74) 
 
 (22.03) 
 
 . . . 
 
OF CAESIUM AND THORIUM. 417 
 
 The salt loses water slowly in the desiccator over sulphuric 
 acid. A sample dried in this way lost six per cent in two days, 
 eleven per cent after one week, and twenty per cent, corre- 
 sponding to practically all the water, after one month. 
 
 The two chlorides that we have obtained are different in 
 type from the potassium salt KC1.2ThCl 4 .18H a O described by 
 Cleve,* and from the ammonium salt 8NH 4 Cl.ThCl 4 .8H 2 O 
 described by Chydenius. f It seems certain that the ammonium 
 salt just mentioned represents a mixture for it is described as 
 a sintered mass made in the dry way. 
 
 * Bulletin, xxi, 118. t Fogg. Ann., cxix, 43. 
 
 27 
 
ON A CAESIUM TELLURIUM FLUORIDE. 
 
 BY H. L. WELLS AND J. M. WILLIS. 
 
 SEVERAL tellurium double fluorides have been described: 
 NaF.TeF 4 by Berzelius, KF.TeF 4 , NH 4 F.TeF 4 , and BaF 2 . 
 2TeF 4 .H 2 O by Hogbom.* It is noticeable that all these 
 fluorides belong to a type which is different from that of the 
 double chlorides, bromides, and iodides of tellurium, e. g., 2KC1. 
 TeCl 4 , 2RbBr.TeBr 4 , and 2CsI.TeI 4 , etc., which have been 
 thoroughly studied in this laboratory by Wheeler, f We have 
 undertaken, therefore, an investigation of the combination of 
 caesium fluoride with tellurium fluoride, with the expectation 
 that possibly several types of double fluorides might be ob- 
 tained. After a systematic examination of the matter, how- 
 ever, we were able to prepare only one double fluoride, 
 CsF.TeF 4 , which corresponds in type to the previously known 
 fluorides. 
 
 A concentrated solution of TeF 4 was prepared by dissolving 
 about 10 g. of pure TeO 2 in an excess of strong hot hydro- 
 fluoric acid, and to this caesium fluoride was added in small 
 portions, the liquid being concentrated by evaporation and 
 cooled after each addition. At the same time small portions 
 of tellurium fluoride were added to a concentrated solution of 
 about 50 g. of caesium fluoride in hydrofluoric acid, and this 
 solution was evaporated and cooled in the same manner. 
 Under the widest range of conditions, however, only a single 
 double salt was obtained. 
 
 1:1 Ccesium Tellurium Fluoride, CsF.TeF 4 . This salt 
 crystallizes beautifully in large, transparent, colorless needles. 
 The presence of free hydrofluoric acid is necessary for its 
 formation, for it is decomposed by water. Several crops, 
 
 * Bulletin, xxv, 60. t Amer. Jour. Sci., Ill, xlv, 267. 
 
ON A CAESIUM TELLURIUM FLUORIDE. 419 
 
 made under widely varying conditions, were analyzed with 
 the following results : 
 
 Calculated for 
 
 
 CsTeF 8 . 
 
 i. 
 
 n. 
 
 IIL 
 
 IV. 
 
 Caesium 
 
 . 37.36 
 
 36.59 
 
 37.50 
 
 38.56 
 
 37.23 
 
 Tellurium . 
 
 . 35.96 
 
 35.51 
 
 36.45 
 
 35.82 
 
 35.60 
 
 Fluorine 
 
 . 26.68 
 
 26.76 
 
 24.51 
 
 26.18 
 
 
 
 Fluorine was determined volumetrically by converting it 
 into SiF 4 , collecting this in water, and titrating with a standard 
 solution of potassium hydroxide. In another portion, after 
 evaporating with concentrated sulphuric acid, and dissolving 
 the residue in hydrochloric acid, tellurium was precipitated with 
 sulphur dioxide, collected on a Gooch crucible, and weighed as 
 metal. From the filtrate from the tellurium caesium was ob- 
 tained, and weighed as normal sulphate. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 May, 1901. 
 
GENERALIZATIONS ON DOUBLE HALOGEN SALTS. 
 
 BY H. L. WELLS. 
 
 THE object of this communication is to point out some con- 
 clusions in regard to double halides in general For this pur- 
 pose a rather full list of these compounds has been prepared. 
 This list is not supposed to be complete, for the literature has 
 not been searched with care except in the cases of certain series 
 which have been studied in the laboratory,* but it is believed 
 to be sufficiently complete for the purpose in view. Care has 
 been taken to consult the original articles where salts of 
 unusual composition have been described, and a few of these 
 have been discarded on account of evidence that they were not 
 real compounds, but no double salt has been rejected simply 
 because it appeared to be irregular. 
 
 It has been thought best to limit the present discussion to 
 the salts of the alkali metals, ammonium, and univalent 
 thallium; for to include the double halides of the organic 
 bases and those in which bivalent metals form the more posi- 
 tive halides would greatly enlarge the list, probably without 
 giving any additional insight into the nature of the compounds. 
 
 For the sake of convenience the double salts will be arranged 
 according to types, which will be designated by ratios indicat- 
 ing the number of atoms of the two metals ; thus, KCl.SnCl 4 
 is a 1 : 1 salt; 2KCl.PtCl 4 is a 2:1 salt; and CsC1.2PbCl a 
 is a 1 : 2 salt. The number referring to the atoms of alkali 
 metal, etc., will be invariably placed first. 
 
 * See Amer. Jour. Sci. from 1892 to the present time. References are 
 given beyond to most of these articles, but reference to general literature is 
 not attempted. 
 
DOUBLE HALOGEN SALTS. 421 
 
 I. SALTS OP UNIVALENT NEGATIVE METALS. 
 
 Cuprous Salts.* 
 
 3:1 2:1 3:2 1:1 
 
 3CsCl.CuCl.H 2 2NH 4 Cl.CuCl 3CsC1.2CuCl NH 4 Cl.CuCl 
 2NH 4 Br.CuBr NH 4 Br.CuBr 
 
 2KCl.CuCl NH 4 I.CuI.H 2 
 
 2:3 1:2 
 
 2NH 4 C1.3CuCl CsCUCuCl 
 
 Silver and Aurous Salts.] 
 
 2:1 1:1 
 
 2KI.AgI NH 4 Cl.AgCl 
 
 2KbI.AgI KCl.AgCl 
 
 2CsCl.AgCl KI.Agl 
 
 KCLAuCl 
 
 II. SALTS OF BIVALENT NEGATIVE METALS. 
 
 Beryllium Salts. 
 
 2:1 1:1 
 
 2KF.BeF 2 KF.BeF 2 
 
 2KCl.BeCl 2 
 
 Magnesium Salts.]. 
 
 1 :1 
 
 NH 4 Cl.MgCl 2 .6H 2 KCl.MgCl 2 .6H 2 
 
 NH 4 Br.MgBr 2 .6H 2 O KBr.MgBr 2 .6H 2 
 
 NH 4 I.MgI 2 .6H 2 KI.MgI 2 .6H 2 
 
 NaF.MgF 2 EbCl.MgCl 2 .6H 2 
 
 NaCl.MgCl 2 .H 2 CsC[MgCl z .QHfl 
 
 CsBr.MgBr 2 .6H 2 
 
 * See Amer. Jour. Sci., HI, xlvii, 96. 
 
 t See Ibid., xliv, 155. 
 
 t See Wells and Campbell, Ibid., Ill, xlvi, 431. 
 
422 
 
 GENERALIZATIONS ON 
 
 Manganese Salts.* 
 
 2:1 
 
 2NH 4 Cl.MnCl 2 .2H 2 
 2EbCl.MnCl 2 .2H 2 O 
 2CsCl.MnCl 2 .2H 2 
 
 1:1 
 
 NH 4 Cl.MnCl 2 .2H 2 O 
 KCl.MnCl 2 .2H 2 
 CsCl.MnCl 2 .2H 2 O 
 
 Ferrous and Nickel Salts. 
 
 2:1 
 
 2NH 4 Cl.FeCl 2 
 
 2KF.FeF 2 
 
 2KCl.FeCl 2 .2H 2 
 
 2NH 4 F.MF 2 .2H 2 
 
 2KF.NiF 2 
 
 1:1 
 
 KF.FeF 2 .H 2 
 
 NH 4 Cl.NiCl 2 .6H 2 O 
 
 NaF.NiF 2 .H 2 
 
 KF.NiF 2 .H 2 
 
 CsCl.NiCl 2 
 
 CsBr.NiBr 2 
 
 3:1 
 
 3CsCl.CoCl 2 
 3CsBr.CoBr a 
 
 Cobalt 
 
 2:1 
 
 2NH 4 F.CoF 2 .2H 2 
 
 2CsCl.CoCl 2 
 
 2CsBr.CoBr 2 
 
 2CsI.CoI 2 
 
 1:1 
 
 NH 4 Cl.CoCl 2 .6H 2 
 NaF.CoF 2 .H 2 
 KF.CoF 2 .H 2 
 CsCl.CoCl 2 .2H 2 
 
 2:1 
 
 2NH 4 F.CuF 2 .2H 2 O 
 2NH 4 Cl.CuCl 2 .2H 2 
 2NH 4 Br.CuBr 2 .2H 2 O 
 2KF.CuF 2 
 2KCl.CuCl 2 
 2CsCl.CuCl 2 
 2CsCl.CuCl 2 .2H 2 O 
 2CsBr.CuBr 2 
 
 Oupric Salts. J 
 
 3:2 
 
 3CsC1.2CuCl 2 .2H 2 
 
 1:1 
 
 NH 4 F.CuF 2 .2H 2 
 
 NH 4 Cl.CuCl 2 .2H 2 
 
 LiCl.CuCl 2 .2H 2 O 
 
 KF.CuF 2 
 
 EbF.CuF 2 
 
 CsCl.CuCl 2 
 
 CsBr.CuBr 2 
 
 * See Saunders, Amer. Chem. Jour., xiv, 152. 
 t See Campbell, Amer. Jour. Sci,, III, xlviii, 418. 
 t See Wells and Dupee, Ibid., Ill, xlvii, 91. 
 
DOUBLE HALOGEN SALTS. 
 
 423 
 
 4:1 
 4NH 4 Cl.ZnCl 2 
 
 1:1 
 
 NH 4 Cl.ZnCl 2 .2H 2 O 
 
 NaF.ZnF, 
 
 NaI.ZnI 2 .3H 2 
 
 KF.ZnF 2 
 
 KI.ZnL, 
 
 Zinc Salts* 
 
 3:1 
 3NH 4 Cl.ZnCl 2 
 
 3CsCLZnCl 2 
 
 3CsBr.ZnBr 2 
 
 3CsI.ZnI 2 
 
 2:1 
 
 2NH 4 F.ZnF 2 .2H 2 
 2NH 4 Cl.ZnCl 2 
 2NH 4 Cl.ZnCl 2 .H 2 
 2NH 4 Br.ZnBr 2 
 2KE 4 I.ZnI 2 
 2NaCl.ZnCl2.3H20 
 2NaI.ZnI 2 .3H 2 
 2KF.ZnF 2 
 2KCl.ZnCl 2 
 2KI.ZnI 3 
 2CsCl.ZnCl 2 
 2CsBr.ZnBr a 
 2CsI.ZnI 2 
 
 4:1 
 
 4NH 4 Cl.CdCl a 
 4NH 4 Br.CdBr 2 
 4KCl.CdCl 2 
 4KBr.CdBr 2 
 
 1:1 
 
 NH 4 F.CdF 2 
 
 NH 4 Br.CdBr 2 4H 2 
 
 JSTaBr.CdBr 2 .2^H 2 
 
 KCl.CdCl, 
 
 KCl.CdCl 2 .iH 2 
 
 KBr.CdBr 2 4H 2 
 
 KI.CdI 2 .H 2 
 
 CsCl.CdCl, 
 
 CsBr.CdBr a 
 
 CsI.CdI 2 
 
 Cadmium 
 
 3:1 
 
 3CsBr.CdBr a 
 3CsI.CdI 2 
 
 2:1 
 
 2NH 4 Cl.CdCl 2 
 2NH 4 Cl.CdCl 2 .H 2 
 
 2NH 4 I.CdI 2 .2H 2 
 
 2NaCl.CdCl 2 .3H 2 
 
 2NaBr.CdBr 2 .5H 2 
 
 2NaI.CdI 2 .6H 2 
 
 2KCl.CdCl 2 
 
 2KCl.CdCl 2 .H 2 
 
 2KBr.CdBr 2 
 
 2KI.CdI 2 .2H 2 
 
 2CsCl.CdCl 2 
 
 2CsBr.CdBr 2 
 
 2CsI.CdI 2 
 
 * See Wells and Campbell, Amer. Jour. Sci., IH, xlvi, 431. 
 t See Wells and Walden, Ibid., 425. 
 
424 
 
 GENERALIZATIONS ON 
 
 4:1 
 4NH 4 Cl.SnCl 2 .3H 2 
 
 2:3 
 
 2NaF.3SnF 2 
 2KF.3SnF 2 .H 2 
 
 Stannous Salts. 
 
 2:1 
 
 2NH 4 F.SnF 2 .2H 2 
 2XH 4 Cl.SnCl 2 .H 2 
 
 1:1 
 
 NH 4 I.SnI 2 
 NH 4 I.SnI 2 .lpI 2 
 NaLSnl, 
 
 3:1 
 
 3CsCl.HgCl 2 
 3CsBr.HgBr 2 
 3CsI.HgI 2 
 
 Mercuric Salts.' 
 
 2:1 
 
 2NH 4 Cl.HgCl 2 .H 2 
 
 2NH 4 Br.HgBr 2 
 
 2NH 4 I.HgI 2 .3H 2 
 
 2NaCl.HgCl 2 
 
 2NaI.HgI 2 
 
 2KCl.HgCl 2 .H 2 
 
 2KBr.HgBr 2 
 
 2KI.HgI 2 
 
 2EbCl.HgCl 2 
 
 2KbCl.HgCl 2 .2H 2 
 
 2CsCl.HgCl 2 
 
 2CsBr.HgBr 2 
 
 2CsI.HgI 2 
 
 2:3 
 
 2NH 4 C1.3HgCl 2 .4H 2 
 2CsI.3HgI 2 
 
 1:2 
 
 KC1.2HgCl 2 .2H 2 
 
 KbC1.2HgCl 2 
 
 CsC1.2HgCl 2 
 
 CsBr.2HgBr 2 
 
 CsI.2HgI 2 
 
 NH 4 Cl.HgCl 2 
 
 NH 4 Cl.HgCl 2 .H 2 
 
 NH 4 Br.HgBr 2 
 
 NH 4 I.HgI 2 .H 2 
 
 LiCl.HgCl 2 
 
 NaCl.HgCl 2 .lH 2 
 
 NaBr.HgBr, 
 
 KCl.HgCl 2 
 
 KCl.HgCl 2 .H 2 
 
 KBr.HgBr, 
 
 KBr.HgBr 2 .H 2 
 
 KI.HgI 2 .liH 2 
 
 KbCl.HgCl, 
 
 CsCl.HgCl 2 
 
 CsBr.HgBr 2 
 
 CsLHgI 2 
 
 TlCl.HgCl 2 
 
 1:6 
 
 NH 4 C1.5HgCl 2 
 CsC1.5HgCl 2 
 
 * See Amer. Jour. Sci., Ill, xliv, 221. 
 
DOUBLE HALOGEN SALTS. 
 
 425 
 
 Palladious, Platinous, and Iridious Salts. 
 
 2:1 
 
 2NH 4 Cl.PdCl 2 
 
 2NH 4 Cl.PtCl 2 
 
 2NHJ.IrI 2 
 
 2TaCl.PdCl 2 
 
 2NaCl.PtCl 2 .4H 2 
 
 4:1 
 
 4CsCl.PbCl f 
 4CsBr.PbBr 2 
 
 1:2 
 
 NH 4 C1.2PbCl 2 
 NH 4 Br.2PbBr 2 
 KC1.2PbCl 2 
 KBr.2PbBr 2 
 KbC1.2PbCl 2 
 B,bBr.2PbBr 2 
 CsC1.2PbCl 2 
 CsBr.2PbBr 2 
 
 2KCl.PdCl 2 
 2KCl.PtCl 2 
 2KBr.PtBr 2 
 2CsCl.PdCl 2 
 2CsCl.PtCl 2 
 
 Lead Salts* 
 
 2:1 
 
 2NH 4 Br.PbBr 2 .H 2 
 2KBr.PbBr 2 .H 2 
 2RbCl.PbCl 2 .H 2 
 2KbBr.PbBr 2 4H 2 
 
 1:1 
 
 LiCl.PtCl 2 .6H 2 
 KCl.PdCl 2 
 KCl.PtCl 2 
 KbCl.PtCl 2 
 
 1:1 
 
 NH 4 Cl.PbCl 2 4H 2 
 
 NH 4 I.PbI 2 .2H 2 
 
 KCl.PbCl 2 .H 2 
 
 KBr.PbBr 2 4H 2 
 
 KBr.PbBr 2 .H 2 
 
 KI.PbI 2 .2H 2 
 
 EbI.PbI 2 .2H 2 O 
 
 CsCl.PbCl 2 
 
 CsBr.PbBr 2 
 
 CsI.PbI 2 
 
 III. SALTS OP TRWALENT NEGATIVE METALS. 
 
 Aluminum Salts. 
 
 3:1 
 
 3KaF.AlF 8 
 3KF.A1F 8 
 
 2:1 
 
 2NaF.AlF 8 
 2KF.A1F 8 
 
 3:2 
 
 3NaF.2AlF 8 f 
 
 3:1 
 
 3NH 4 F.VF 8 
 
 1:1 
 
 NH 4 F.VF 8 .2H 2 
 
 Vanadium Salts. 
 
 5:2 
 
 5NaF.2VF 8 .H 2 
 
 1:1 
 
 NaCl.AlCl 8 
 KC1.A1C1 8 
 KBr.AlBr, 
 KI.A1I 8 
 
 2:1 
 
 2NH 4 F.VF 8 .H 2 
 2KF.VF 8 .H 2 
 
 * Nearly all the older work on double halides of lead was entirely erro- 
 neous. See Remsen and Herty, Amer. Chem. Jour., xiv, 107 ; Wells, Amer. 
 Jour. Sci., Ill, xlr, 121 ; Wells and Johnson, Ibid., xlvi, 25, 34. 
 
 t The formula for chiolite is somewhat doubtful. Some authorities prefer 
 the formula 6NaF.3AlF 8 , but the simpler formula is selected here. 
 
426 GENERALIZATIONS ON 
 
 Chromium Salts.* 
 
 3:1 2:1 1:1 
 
 3NH 4 F.CrF 8 2NH 4 F.CrF 8 .H 2 KCl.CrCl 8 
 
 3KF.CrF 8 2NH 4 Cl.CrCl 3 .H 2 
 
 3KCl.CrCl 8 2NaF.CrF 8 .H 2 
 
 3TlCl.CrCl 8 2KF. CrF 3 .H 2 O 
 
 2KCl.CrCl 3 .2H 2 
 
 2CsCl.CrCl 3 .H 2 
 
 2CsCl.CrCl 8 .4H 2 
 
 Titanous and Manganic Salts. 
 
 3:1 2:1 
 
 3NH 4 F.TiF 8 2NH 4 F. MnF 8 
 
 2NaF.MnF 8 
 2KF.MnF 8 
 2NH 4 F.TiF 8 
 
 Ferric SaltsJ 
 
 3:1 2:1 1:1 
 
 3NH 4 F.FeF 8 2NH 4 F.FeF 3 NH 4 Br.FeBr 8 .2H 2 
 
 3KF.FeF 8 2NH 4 Cl.FeCl 8 CsCl.FeCl s 
 
 3CsCl.FeCl 8 .H 2 2NaF.FeF 8 .|H 2 CsBr.FeBr a 
 3TlCl.FeCl 8 2NaCl.FeCl 8 .H 2 
 
 2KF.FeF 8 
 
 2KCl.FeCl 8 .H 2 
 
 2BbCl.FeCl 3 H 2 
 
 2EbBr.FeBr s .H 2 
 
 2CsCl.FeCl 8 .H 2 
 
 2CsBr.FeBr 8 .H 2 O 
 
 Arsenious Salts.$ 
 
 3:2 
 
 3EbC1.2AsCl 8 3CsC1.2AsCl 8 
 
 3EbBr.2AsBr 8 3CsBr.2AsBr 3 
 
 3RbI.2AsI 8 3CsI.2AsI 8 
 
 * See Wells and Boltwood, Amer. Jour. Sci., III, i, 249. 
 t See Walden, Ibid., Ill, xlviii, 283. 
 J Wheeler, Ibid., Ill, xlvi, 88. 
 
DOUBLE HALOGEN SALTS. 
 
 427 
 
 Indium Salts. 
 
 8:1 
 
 3KCl.InCl 8 .lpI 2 
 
 2:1 
 2NH 4 Cl.InCl 8 .H 2 
 
 4:1 
 4NH 4 I.SbI 8 .3H 2 O 
 
 2:1 
 
 2NH 4 F.SbF 8 
 
 2NH 4 Cl.SbCl 8 
 
 2NH 4 Cl.SbCl 8 .H 2 O 
 
 2LiF.SbF 8 
 
 2KF.SbF 3 
 
 2KCl.SbCl 8 
 
 2KCl.SbCl 8 .2H 2 
 
 2CsF.SbF 8 
 
 Antimonious Salts. 
 
 3:1 
 
 3NH 4 Cl.SbCl 8 .liH 2 O 
 3NaF.SbF 8 
 
 SNaCLSbCL 
 
 3:2 
 
 3NH 4 L2SbI 8 .3H 2 
 
 3NaI.2SbI 8 .12H 2 
 
 3KI.SbI 8 .3H 2 
 
 3EbC1.2SbCl 8 
 
 3EbBr.2SbBr 8 
 
 3RbI.2SbI 8 
 
 3CsC1.2SbCl 8 
 
 3CsI.2SbI 
 
 3:4 4:7 
 
 3NH 4 I.4SbI 8 .9H 2 4CsF.7SbF 8 
 
 1:3 
 
 CsF.3SbF 8 
 
 7:3 
 
 7KC1.3SbCl 8 
 7KBr.3SbBr 8 .8H 2 
 7RbC1.3SbCl 8 
 7EbBr.3SbBr 8 
 
 1:1 
 
 NH 4 Cl.SbCl 3 
 NH 4 I.SbI 8 .2H 2 O 
 KF.SbF 8 
 KI.SbI 8 .H 2 
 
 EbCl.SbCl 8 
 CsF.SbF 8 
 
 1:2 
 
 EbC1.2SbCl 8 
 CsF.2SbF 8 
 
 Auric Salts, f 
 
 1:1 
 
 NH 4 Cl.AuCl 8 .H 2 
 
 NaCl.AuCl 8 .2H 2 O 
 
 NaBr.AuBr 8 .2H 2 O 
 
 KCl.AuCl 8 4H 2 
 
 KCl.AuCl 8 .2H 2 O 
 
 KBr.Aubr 8 .2H 2 O 
 
 KIAuI 8 
 
 EbCl.AuCl, 
 
 RbBr.AuCl 8 
 
 CsCl.AuCl 8 
 
 CsCl.AuCl 8 4H 2 O 
 
 CsBr.AuBr 8 
 
 TlCl.AuCl 8 
 
 * See Wheeler, Amer. Jour. ScL, in, Ixvi, 269; Wells and Metzger, Ibid., 
 June, 1901. 
 
 t See Wells and Wheeler, Ibid., HI, xliv, 157. 
 
428 
 
 GENERALIZATIONS ON 
 
 4:1 
 
 4NH 4 Cl.BiCl 8 
 4KI.BiI 8 
 
 2:1 
 
 2NH 4 Cl.BiCl 8 
 2NaCl.BiCl 8 .3H 2 
 2KCl.BiCl 3 .2H 2 
 2KI.BiI 8 .H 2 O 
 
 1:2 
 
 NH 4 C1.2BiCl 8 . 
 
 Bismuth Salts. 
 
 3:1 
 
 3NH 4 Cl.BiCl 8 
 
 3NaCl.BiCl 8 
 
 3KCl.BiCl 8 
 
 3KI.BiCl 8 
 
 3EbCl.BiCl 8 
 
 3CsCl.Bi01 8 
 
 3:2 
 
 3NaI.2BiI 8 
 3KC1.2BiCl 8 
 3CsC1.2BiCl 8 
 
 7-3 
 7EbC1.3BiCl 8 
 
 NH 4 Cl.BiCl 8 
 KCl.BiCl 8 .H 2 
 KI.BiI 8 
 KbCl.BiCl 8 .H 2 
 
 5:1 
 5T1I.T1I 8 
 
 3:2 
 
 3KC1.2T1C1 8 .1H 2 
 3LBr.2TlBr 8 .3H 2 
 3KI.2T1I 8 .3H 2 
 3CsC1.2T101 8 
 3CsBr.2TlBr 8 
 
 Thallic Salts* 
 
 3:1 
 
 3NH 4 C1.T1C1 8 .2H 2 O. 
 3NH 4 C1.T1C1 8 
 3NH 4 I.T1I 8 
 3LiCl.TlCl 8 .8H 2 
 3NaCl.TlCl 8 .12H 2 
 3KC1.T1C1 S .2H 2 
 3KbCl.T101 8 .H 2 
 3EbBr.TlBr 8 .H 2 
 3CsCl.TlCl 8 .H 2 
 3T1C1.T1C1 8 
 3TlBr.TlBr 8 
 
 2:1 
 
 2KC1.T1C1 8 .2H 2 
 2EbCl.TlCl 8 .H 2 
 2CsCl.TlCl 8 
 2CsCl.TlCl 8 .H 2 
 
 1:1 
 
 NH 4 Br.TlBr 8 .2H 2 
 NH 4 Br.TlBr 8 .4H 2 
 NH 4 Br.TlBr 8 
 NH 4 I.T1I 8 
 KBr.TlBr, 
 
 KI.TlIg 
 
 EbBr.TlBr 8 
 
 EbI.TlI 3 
 
 CsBr.TlBr 8 
 
 CsI.TlI 8 
 
 T1C1.T1C1 8 
 
 TlBr.TlBr 8 
 
 * See Pratt, Amer. Jour. Sci., Ill, xlix, 397. 
 
DOUBLE HALOGEN SALTS. 
 
 429 
 
 Rhodium, Ruthenium, Iridium, and Osmium Salts. 
 
 3:1 
 
 3NH 4 Cl.RhCl 8 .HH 2 3NaBr.IrBr 8 4H 2 O 
 3NH 4 Cl.IrCl 3 .liH 2 3KCl.EhCl 8 
 3NH 4 Br.IrBr 8 4H 2 3KCUrCl 8 .3H 2 
 3NH 4 I.IrI 8 4H 2 3KBr.IrBr 8 .3H 2 
 3NaCl.RhCl 8 .9H 2 SKLIrl, 
 3NaCl.IrCl 8 .12H 2 3KC1.0sCl 3 .6H 2 
 
 2:1 
 
 2NH 4 Cl.KliCl 8 .H 2 
 2NH 4 Cl.RuCl 8 
 2NH 4 Cl.OsCl 8 .liH 2 
 2KCl.KhCl 8 .H 2 
 2KCl.EhCl 8 .3H 2 
 2KCl.R,uCl 8 
 
 IV. SALTS OF QUADKIVALENT NEGATIVE METALS. 
 Titanic,* Germanium, and Manganese Salts. 
 
 2:1 
 
 2NH 4 F.TiF 4 2KbF.TiF 4 
 
 2NaF.TiF 4 2KF.MnF 4 
 
 2KF.TiF 4 .H 2 2KF.GeF 4 
 
 4:1 
 
 4LiF.ZrF 4 .H 2 
 
 6:3 
 5TlF.3ZrF 4 
 
 Zirconium Salts, f 
 
 3:1 
 
 3NH 4 F.ZrF 4 
 3KF.ZrF 4 
 
 5:2 
 5NaF.2ZrF 4 
 
 1:1 
 
 KF.ZrF 4 .H 2 
 CsF.ZrF 4 
 TlF.ZrF 4 
 TlF.ZrF 4 .H 2 
 
 2:1 
 
 2NH 4 F.ZrF 4 
 2LiF.ZrF 4 
 2NaF.ZrF 4 
 
 2NaGl.ZrGl 4 
 
 2KF.ZrF 4 
 2CsF.ZrF 4 
 
 2:3 
 2CsF.3ZrF 4 .2H 2 
 
 * Rose's compounds, 6NH 4 Cl.TiCl 4 and 3NH 4 Cl.TiCl 4 are omitted here 
 because they represent sublimates of variable composition. Pennington's 
 salt 4CsF.TtF 4 is also left out, as it was described with two other caesium 
 salts of very suspicious composition which are referred to under the pen- 
 tivalent compounds. 
 
 t See Wells and Foote, Amer. Jour. ScL, IV, i, 18; iii, 466. 
 
430 
 
 GENERALIZATIONS ON 
 
 Stannic and Antimony Salts. 
 
 4:1 2:1 
 
 4NH 4 F.SnF 4 2NH 4 Cl.SnCl 4 2KF.SnF 4 .H 2 
 
 2LiF.SnF 4 .2H 2 O 2KCl.Sn01 4 
 
 2NaF.SnF 4 2KbCl.SnCl 4 
 
 2NaCl.SnCl 4 2CsCl.SnCl 4 
 
 2NaCl.SnCl 4 .6H a O 2CsCl.SbCl 4 
 23TaBr.SnBr 4 .6H 2 
 
 2KCl.TeCl 4 
 
 2KBr.TeBr 4 .2H 2 
 
 2KI.TeI 4 .2H 2 
 
 2RbCl.TeCl 4 
 
 2EbBr.TeBr 4 
 
 Tellurium Salts.* 
 
 2EbI.TeI 4 
 2CsCl.TeCl 4 
 2CsBr.TeBr 4 
 2CsI.TeI 4 
 
 Ceric Salt. 
 
 3:2 
 
 3KF.CeF 4 .2H 2 
 
 NH 4 F.TeF 4 
 
 KF.TeF 4 
 
 CsF.TeF 4 
 
 2NH 4 Cl.PdCl 4 
 
 2NH 4 Cl.RuCl 4 
 
 2NH 4 Cl.IrCl 4 
 
 2NH 4 Cl.Pt01 4 
 
 2KH 4 C1.0sCl 4 
 
 2NH 4 Br.PtBr 4 
 
 2NH 4 Br.IrBr 4 
 
 2NH 4 I.PtI 4 
 
 2LiCl.PtCl 4 .6H 2 
 
 2NaCl.PtCl 4 .6H 2 
 
 Platinum Group Salts. 
 
 2: 1 
 
 2NaCl.IrCl 4 .6H 8 
 
 2NaC1.0sCl 4 
 
 2NaBr.IrBr 4 
 
 2NaI.PtI 4 .6H 2 
 
 2NaI.IrI 4 
 
 2KCl.PdCl 4 
 
 2KCl.KuCl 4 
 
 2KCl.PtCl 4 
 
 2KCl.IrCl 4 
 
 2KC1.0sCl 4 
 
 2KBr.PdBr 4 
 
 2KBr.PtBr 4 
 
 2KBr.IrBr 4 
 
 2KI.PtI 4 
 
 2KLM 4 
 
 2EbCl.PtCl 4 
 
 2CsCl.PtCl 4 
 
 2TlCl.PtCl 4 
 
 See Wheeler, Amer. Jour. Sci., in, xlv, 267. 
 
DOUBLE HALOGEN SALTS. 431 
 
 Plumbic Salts* 
 
 2:1 
 
 2NH 4 Cl.PbCl 4 2RbCl.Pb01 4 
 
 2KCl.PbCl 4 2CsCl.Pb01 4 
 
 Thorium Salts, f 
 
 3:1 2:1 
 
 3CsCl.ThCl 4 .12H 2 2KF.ThF 4 .4H 2 
 
 2CsCl.ThCl 4 .llH 2 O 
 
 1:1 1:2 
 
 KF.ThF 4 4H 2 KC1.2TliCl 4 .18H 2 
 
 V. SALTS OF QUINQUIVALENT NEGATIVE METALS.J 
 
 4:1 2:1 1:1 
 
 4NH 4 Cl.SbCl 6 2KF.AsF 5 KF.AsF 6 
 
 2NH 4 F.SbF 6 NH 4 F.SbF 6 
 
 2KF.SbF 5 NaF.SbF 5 
 
 2KF.NbF 6 KF.SbF 6 
 
 2EbF.NbF 6 
 2NH 4 F.TaF 6 
 2NaF.TaF 6 
 2KF.TaF 6 
 2EbF.TaF 6 
 
 * See Amer. Jour. ScL, III, xlvi, 180. 
 
 t The salt SNH^CLThCl^SHaO, described by Chydenius as a sintered 
 mass is rejected, as is also 8KF.7ThF 4 .6H 2 O, which Chydenius himself con- 
 sidered to be probably a mixture. 
 
 t Pennington's salts, 7CsF.NbF 5 and 16CsF.TaF 6 are omitted because they 
 differ entirely from two rubidium salts obtained in the same investigation 
 (which are given in the list), and because they depend upon single partial 
 analyses. (See Jour. Amer. Chem. Soc., xviii, 69.) It is impossible to 
 believe that caesium and rubidium salts of the same metals should differ so 
 widely as they appear to do. 
 
432 
 
 GENERALIZATIONS ON 
 
 10 
 
 ..<N 
 
 | : - 
 
 rj CO * CO 
 
 4 
 
 * 
 
 3 
 
 r-\ 
 t- 
 
 ft 
 02 
 
 CO CO 
 
 CO 
 
 (M 
 
 CO 
 <M 
 
 CO 
 
 CO 
 
 I 
 
 I 
 
 -2 
 
 O 
 
 I 
 
 1 
 
 02 
 
 CO 
 
 CO 
 
 CO 
 
 co" co 
 
 CO 
 XO 
 
 CO CO 
 
 CO 
 
 CO 
 
 1O 
 
 CO CO 
 
 CO 
 
 10 
 
DOUBLE HALOGEN SALTS. 433 
 
 Upon examining this table a marked similarity is to be 
 noticed in the different series ; the 2 : 1 and 1 : 1 types are 
 common to all ; the 4 : 1, 3 : 1, 3 : 2, and 1 : 2 types are found 
 in all but one ; while 2 : 3 types are found in three of the 
 series. Besides the seven types just mentioned there are eight 
 others, but only one of the latter exists in more than one 
 series, and most of them are represented by only a single salt. 
 It is quite possible that some of these unusual ratios are due 
 to erroneous descriptions of salts, but it is certain that many 
 of them represent real compounds. 
 
 The remarkable similarity in the prominent types of the 
 series of different valencies leads to the conclusion that the 
 valency of a negative Tialide has no influence upon the types of 
 double salts that it forms. For, if valency had an influence, it 
 would be expected that the five series would show marked 
 differences from one another, and probably the halides of 
 higher valency would tend to combine with a larger number 
 of alkaline halide molecules. Instead of progression of types 
 as valency increases, however, the tables show a marked 
 symmetry in types, and their arithmetical limits in both direc- 
 tions are as nearly constant as could be expected considering 
 the variations in the numbers of known salts in the different 
 series. The prominence of the 3 : 1 and 3 : 2 types in the 
 trivalent series may perhaps be taken as an indication that the 
 rule is not absolute, but since these types are not the only 
 prominent ones, and since both of them occur in three other 
 series, the circumstance that many are found in the trivalent 
 series may be accidental. 
 
 The facts, mentioned above, that the ratios of the different 
 series are nearly symmetrical and that the arithmetical limits 
 of the types in loth directions are nearly uniform appear to be 
 important, for they indicate that the ratios according to which 
 positive and negative halides combine are not influenced to 
 any great degree by the positive or negative nature of the 
 halides. In other words, the molecules of alkaline halides 
 possess nearly the same combining power as molecules of negative 
 halides. Perhaps this point may be made clearer by the state- 
 
 28 
 
434 GENERALIZATIONS ON 
 
 ment that if all the ratios in the table are read backward they 
 remain almost unchanged in their arithmetical aspect, and in 
 the cases of most of the prominent ratios there is actually no 
 change ; for instance, the list of ratios 5 : 1, 3 : 1, 2 : 1, 3 : 2, 
 1 : 1, 2 : 3, 1 : 2, 1 : 3 and 1 : 5 is the same whether read for- 
 ward or backward, and to these ratios belong almost 95 per 
 cent of the salts here considered. It is evident, however, that 
 positive and negative halides are not of exactly equal numerical 
 importance, for to the left of the 1 : 1 ratio in the table, toward 
 the positive halide end, the ratios are more numerous and the 
 salts far more abundant than toward the negative halide end ; 
 it appears to be the 3 : 2 ratio column that forms a nearly 
 symmetrical dividing line between the positive and negative 
 ends, and even with this division the salts on the positive side 
 predominate in number. 
 
 Another point to which attention may be called is that salts 
 of simple types predominate. More than 71 per cent of the 
 salts in the list belong to the 2 : 1 and 1 : 1 ratios, while the 
 4 : 1, 3 : 1, 3 : 2, 2 : 3, and 1 : 2 ratios represent over 25 per 
 cent of them. The remaining eight types, most of which are 
 more complex, include less than three per cent of all the 
 salts. The fact that the 2 : 1 and 1 : 1 ratios are so important 
 in all the series is another indication that valency does not 
 influence the ratios according to which halides combine. 
 Another evidence of simplicity is the circumstance that as far 
 as is known not more than five molecules of one halide can 
 combine with one molecule of another in extreme cases, while 
 the usual limit is 2, 3, or 4. It is undoubtedly a fact that 
 there are a few complicated salts which are not derived from 
 mixtures or poor analyses. For example, the antimony salts 
 to which is here given the 7 : 3 ratio have been thoroughly 
 investigated, and without doubt possess either this ratio or one 
 still more complex. 
 
 The conclusions that have been reached above are not 
 encouraging in the way of giving an insight to the structure 
 of double halides, for, if valency plays no part in this, the 
 number of halogen atoms likewise has no influence, and such 
 
DOUBLE HALOGEN SALTS. 435 
 
 laws or rules, based on halogen atoms, as have been advanced 
 must be abandoned. Professor Remsen's law,* which states 
 that the number of alkaline halide molecules that can combine 
 with a negative halide molecule is not greater than the valency 
 of the latter, is one of these. The stepped line in the table of 
 ratios shows a region in the upper left-hand corner which is 
 beyond the domain of this law. At the tune that the law was 
 brought forward it seemed reasonable, for there were only two 
 or three possible exceptions to it. With our present knowl- 
 edge of the subject, however, it is evident that it is an acci- 
 dental circumstance that the law applies to so many double 
 halides, for it is the small number of known double halides of 
 univalent negative metals, as well as the rarity of salts of high 
 alkali-metal ratios in the subsequent series, which determines 
 the comparatively small number of exceptions to this rule. 
 Werner's theory,! depending as it does upon definite numbers 
 of halogen and other atoms, appears to have no application to 
 the double halides in general. Possibly it may apply to the 
 salts that form complex ions in solution. 
 
 A prevalent idea concerning the influence of valency in the 
 formation of simple oxygen salts seems to need revision also, 
 for halogen and oxygen salts appear to be governed by the 
 same laws. For instance, there is no good reason for consid- 
 ering the valency of sulphur to have an influence upon the 
 types of sulphates that can exist, and consequently for imagin- 
 ing an ideal or " ortho " sulphate such as K 6 SO 6 . 
 
 As far as double oxygen salts are concerned, there can be 
 little doubt that they are analogous to double halogen salts. 
 Their molecules usually unite in 2 : 1 or 1 : 1 ratios. A 
 double oxygen salt, however, in view of the analogy between 
 double halogen and simple oxygen salts, may be considered as 
 analogous to a double-double halogen salt, if such there be. 
 Possibly some of the complex types of double halogen salts 
 may be due to such combinations; for instance, 2Rb a SbCl 6 
 (unknown) + RbSbCl 4 = 7RbC1.3SbCl 3 . 
 
 * Amer. Chem. Jour., xi, 296 ; xiv, 85. 
 t Zeitschr. anorg. Chem., iii, 267. 
 
436 GENERALIZATIONS ON 
 
 The stability of double halogen salts when they are dissolved 
 in water varies exceedingly, but since gradations occur, and 
 there is no sharp dividing line between classes of different 
 stability, it cannot be supposed that there is any real difference 
 in structure, as far as the solid state is concerned, between 
 these different classes. In fact, salts that are isomorphous and 
 evidently strictly analogous vary greatly in their behavior in 
 solution ; for example, the octahedral salts K 2 PtCl 6 , KaSnCle, 
 and KaPbCls. Although only two classes of double salts, based 
 on their behavior in solution, are usually recognized, there are 
 three more or less distinctly denned groups. The first, of 
 which K 8 PtCl 6 is an example, undergo ionization into alkali- 
 metal ions and complex negative ions with little or no de- 
 composition into the simple halides. It is this class which 
 displays a striking similarity to simple oxygen salts. They 
 are comparatively few in number, but the double cyanides, 
 which are probably entirely analogous to double halides, fur- 
 nish many examples of them. The second group consists of 
 salts that readily separate into their component halides in solu- 
 tion, but which may be recrystallized unchanged from water or 
 from dilute acid solutions. To the third class belong salts 
 that require the presence of an excess of one of their halides 
 in solution in order that they may be formed. Sometimes the 
 excess required is very great, so that the proportion of halides 
 in the solution is widely different from that in the salt that 
 crystallizes out. It is by varying the proportions of two 
 halides greatly that a series of several double salts may some- 
 times be produced from them. 
 
 The fact that the same two halides may unite in several 
 proportions to form different double salts is a point that has 
 been sometimes disregarded by chemists. Some investigators 
 have been content with the preparation of a single compound 
 in cases where several might have been produced. This is a 
 matter that should be borne in mind in the study of the physi- 
 cal properties of solutions. The largest number of double 
 halides that have been produced by combining two simple 
 halides appears to be five ; for instance, five csesium mercuric 
 
DOUBLE HALOGEN SALTS. 437 
 
 chlorides and five caesium antimonious fluorides, are the most 
 extensive series of this kind that are known. Such series 
 appear to be limited by a sort of overlapping of conditions ; 
 for compounds that are expected from analogy, intermediate 
 between two that are produced, often cannot be prepared, and 
 where a single double salt is very stable it may be the only 
 one that can be made. Where several double halides of the 
 same two simple halides exist, it is usually the case that one of 
 them is the most stable one, and that this is produced by the 
 recrystallization of the others, but cases occur where two of 
 the salts, at least, are unchanged by recrystallizing. 
 
 The water of crystallization of double halides appears to 
 present precisely the same problem as in the case of simple 
 salts, for we find double salts of the same type possessing no 
 water and varying quantities of it, and no good evidence is to 
 be found that in these compounds a molecule of water is 
 equivalent to an alkaline halide molecule, a relation similar to 
 that, for example, that has been supposed to exist between the 
 compounds FeSO 4 .7H 2 O and FeSO 4 .(NH 4 ) 2 SO 4 .6H 2 O. 
 
 It has been pointed out previously * that double halides 
 seem to show an increase in stability and variety from the 
 iodides to the chlorides, and apparently also to the fluorides. 
 It is probable that this is a general rule among these compounds. 
 
 It has been shown f also that with the metals magnesium, 
 zinc, cadmium, and mercury the tendency to form double 
 halides increases with the atomic weights. This tendency is 
 found in other cases also, but from a comparison of the zir- 
 conium and thorium salts, as well as the antimonious and 
 bismuth salts, it seems doubtful that this is a rule that applies 
 in all cases. 
 
 It has been found in some cases J that caesium appears to 
 form more extensive series of double salts than the other 
 alkali-metals, but evidently there are exceptions to this rule, 
 for Wells and Campbell were unable to prepare any caesium 
 magnesium iodide, although both ammonium and potassium 
 magnesium iodides are known. 
 
 * Amer. Jour. Sci., Ill, xlvi, 431. t Ibid., xlvi, 434. 
 
 } Ibid., Ill, xlvi, 223. Ibid., 432. 
 
438 GENERALIZATIONS ON 
 
 The rule advanced by Godeffroy * that all the double salts 
 of caesium are less soluble than those of the other alkali 
 metals, while true in most instances, is apparently not invariable, 
 for Wells and Campbell f found that the 1 : 1 caesium-zinc 
 salts, if they exist at all, are too soluble to be crystallized in a 
 satisfactory condition, although corresponding ammonium and 
 potassium salts are known. 
 
 Professor Remsen J has called attention to the fact that cer- 
 tain double halides show gradations in water of crystallization, 
 increasing with the atomic weight of the halogen and decreas- 
 ing with the atomic weight of the alkali metal. An inspection 
 of the lists of salts shows that there are many instances of 
 analogous salts to which these rules apply, particularly in the 
 larger amounts of water in sodium and lithium salts and in 
 iodides, and the smaller amounts of water in caesium salts 
 and in fluorides; but, taking the list as a whole, these gra- 
 dations are not very striking, and there are a few apparent 
 exceptions to them. 
 
 It has been noticed in some cases that caesium halides com- 
 bine with a greater number of negative halide molecules than 
 do other alkaline halides, while at the other end of a series 
 more molecules of a lighter alkaline halide than of a caesium 
 halide may combine with a negative halide. For instance, the 
 caesium-zirconium and potassium-zirconium fluorides show 
 this relation in the following ratios of their salts : 
 
 CsF : ZrF 4 2:1 1:1 2:3 
 
 KF:ZrF 4 ... 3:1 2:1 1:1 
 
 This is evidently not a general rule, however, for there is a 
 1 : 5 ammonium-mercuric chloride corresponding to the extreme 
 type of caesium-mercuric chlorides, while at the positive end 
 the caesium-lead and caesium-mercuric salts extend farther than 
 the salts of other alkali metals. 
 
 * Berichte, ix, 1365. t Amer. Jour Sci., Ill, xlvi, 434. 
 
 t Amer. Chem. Jour., xiv, 88. 
 
 Wells and Foote, Amer. Jour. Sci., IV, i, 18. 
 
DOUBLE HALOGEN SALTS. 439 
 
 Many irregularities appear in the list. Sometimes fluorides, 
 chlorides, bromides, and iodides seem to be analogous, but often 
 they are not. There are some curious relations of this sort. 
 For instance, all the double fluorides of tellurium are 1 : 1 
 salts, while the other tellurium salts belong to the 2 : 1 type ; 
 there is only one chloride among twelve 1 : 1 thallic salts, 
 while all the 2 : 1 salts and nearly all the 3 : 1 salts are 
 chlorides. 
 
 Perhaps the most marked case of uniformity in double 
 halides is the invariability of the 2 : 1 type in the salts of the 
 quadrivalent platinum group metals, and this type predom- 
 inates to a remarkable extent among the salts of the other 
 quadrivalent elements. There are several other well-marked 
 groups, such as the 1 : 1 magnesium salts with six molecules 
 of water, the 2 : 1 chromium and ferric salts with one water, 
 and the 3 : 2 anhydrous arsenious, antimonious, and bismuth 
 salts. 
 
 It may be stated that in our study of double halides about 
 one-third of the compounds given in the list have been pre- 
 pared in this laboratory. Some indications of regularity have 
 been observed in connection with these researches, but it 
 must be admitted that the results have been negative as far as 
 throwing light upon the structure of this class of compounds 
 is concerned. 
 
 Summary. In the present discussion, by the method of 
 comparison, the most important conclusion appears to be that 
 the valency of negative halides has little or no influence upon 
 the types of double halides that they form. 
 
 It has been shown also that the combining power of negative 
 halides, whatever their valency may be, is nearly the same as 
 that of alkaline halides. 
 
 Attention has been called to the prominence of simple types 
 among these double salts. 
 
 It has been pointed out that double halides probably increase 
 in ease of formation and variety from the iodides to the fluo- 
 rides ; but that other gradations and analogies which exist in 
 some cases are probably not general. 
 
440 DOUBLE HALOGEN SALTS. 
 
 The classification of double halides into three groups based 
 upon their behavior in solution has been advocated: (1) 
 Salts that form complex ions. (2) Other salts that can be 
 recrystallized from water or dilute acids. (3) Salts that 
 require the presence of an excess of one of their components 
 for their formation. 
 
 SHEFFIELD SCIENTIFIC SCHOOL, 
 May, 1901. 
 
INDEX 
 
INDEX 
 
 ALUMS, gradations in the properties of, 
 
 170, 190. 
 
 Alums, solubilities of, 170, 190. 
 Ammonium-cuprous halides, 385. 
 Ammonium-ferric halides, 357. 
 Ammonium-lead halides, 283, 313. 
 Ammonium nitrates, acid, 146. 
 Autimonic double halides, 413. 
 Antimony, a salt of quadrivalent, 139. 
 Antimony double halides, 139, 320,394, 
 
 407,413. 
 Arsenic double halides, 300. 
 
 BISMUTH double halides, 397. 
 
 CADMIUM double halides, 334. 
 Caesium-antimonic halides, 413. 
 Caesium-antimonious halides, 407. 
 Caesium-antimony chloride, quadriva- 
 lent, 139. 
 
 Caesium-arsenious halides, 300. 
 Caesium-bismuth halides, 397. 
 Caesium-bismuth nitrate, 153. 
 Caesium-cadmium halides, 334. 
 Caesium-chromium halides, 381 . 
 Caesium-cobalt halides, 366. 
 Caesium-cupric halides, 347, 352. 
 Caesium-cuprous halides, 354. 
 Caesium-ferric halides, 357. 
 Caesium-ferric nitrate, 152. 
 Caesium-gold halides, 211. 
 Caesium iodate-periodate, 155. 
 Caesium iodates, 58. 
 Caesium-lead halides, 250, 313. 
 Caesium-magnesium halides, 342. 
 Caesium material, purification of, 142. 
 Caesium-mercuric halides, 218. 
 Caesium-nickel halides, 366. 
 Caesium nitrates, acid, 146. 
 
 Caesium pentahalides, 48. 
 Caesium periodate, 155. 
 Caesium, preparation of pure salts of, 71. 
 Caesium, quantitative determination, 71. 
 Caesium-silver halides, 30, 207. 
 Caesium-tellurium fluoride, 418. 
 Caesium-tellurium halides, 268. 
 Csesium-thallic halides, 370. 
 Caesium-thorium chlorides, 415. 
 Caesium trihalides, 10. 
 Caesium-uranyl chloride, 384. 
 Caesium-zinc halides, 342. 
 Caesium-zirconium fluorides, 390. 
 Chromium double halides, 381. 
 Cobalt double halides, 366. 
 Copper double halides, 347, 352, 354, 
 
 385. 
 
 Cupric double halides, 347, 352. 
 Cuprous double halides, 354, 385. 
 
 FERRIC double halides, 357. 
 Ferrous-ferric bromides, 364. 
 
 GENERALIZATIONS on double halides, 
 
 420. 
 Gold double halides, 211. 
 
 INORGANIC compounds, properties of, 
 
 158. 
 
 Iodates, alkali-metal, 58. 
 Iron double halides, 357. 
 Iron, volumetric determination of, 97. 
 
 LEAD compounds with extra iodine, 77, 
 
 89. 
 Lead double halides, 250, 283, 295, 313, 
 
 77. 
 
 Lead tetrachloride double salts, 313. 
 Lithium peutahalide, 48. 
 
444 
 
 INDEX 
 
 Lithium-thallic halides, 370. 
 Lithium-zirconium fluorides, 400. 
 
 MAGNESIUM double halides, 342. 
 Mercuric double halides, 218. 
 
 NICKEL double halides, 366. 
 Nitrates, acid, 146. 
 Nitrates, double, 151. 
 
 PECULIAR halides of potassium and 
 
 lead, 77. 
 
 Pentahalides, alkali-metal, 48. 
 Periodic system, 158. 
 Potassium-arsenious halides, 300. 
 Potassium-ferric halides, 357. 
 Potassium ferricyanide, an isomer of, 
 
 116, 130. 
 
 Potassium ferrous-ferric bromide, 369. 
 Potassium iodate-chloride, 68. 
 Potassium-lead halides, 250, 313, 77. 
 Potassium-lead halides, peculiar, 77. 
 Potassium nitrate, acid, 146. 
 Potassium pentahalide, 48. 
 Potassium-silver halides, 207. 
 Potassium-tellurium halides, 268. 
 Potassium trihalides, 33. 
 
 RuBiDiUM-antimonious halides, 320, 
 
 394. 
 
 Rubidium-antimony oxychloride, 397. 
 Rubidium-arsenious halides, 300. 
 Rubidium-ferric halides, 357. 
 Rubidium ferrous-ferric bromide, 364. 
 
 Rubidium-gold halides, 211. 
 Rubidium iodates, 58. 
 Rubidium-lead halides, 295, 313. 
 Rubidium nitrates, acid, 146. 
 Rubidium pentahalide, 48. 
 Rubidium-silver halides, 207. 
 Rubidium-tellurium halides, 268. 
 Rubidium-thallic halides, 370. 
 Rubidium trihalides, 33. 
 
 SILICIC acid, separation from tungstic 
 
 acid, 136. 
 
 Silver double halides, 207. 
 Sodium pentahalide, 48. 
 Sodium-thallic halides, 370. 
 Sodium-zirconium fluorides, 400. 
 
 TELLURIUM double halides, 268, 418. 
 Thallic double halides, 370. 
 Thallium triiodide, 84. 
 Thallium-zirconium fluorides, 400. 
 Thallous nitrates, acid, 146. 
 Thallous-thallic nitrate, 154. 
 Thorium double chlorides, 415. 
 Titanic acid, volumetric determination 
 
 of, 97. 
 
 Trihalides, alkali-metal, 10, 33. 
 Tungstic acid, separation from silicic 
 
 acid, 136. 
 
 VANADIUM, compounds of trivalent, 
 103. 
 
 ZINC double halides, 342. 
 Zirconium double fluorides, 390, 400. 
 
 RAR^ 
 
 OF THE 
 
 UNIVERSITY 
 
\LIFORF 
 
 THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 AN INITIAL FINE OF 25 CENTS 
 
 WILL BE ASSESSED FOR FAILURE TO RETURN 
 THIS BOOK ON THE DATE DUE. THE PENALTY 
 WILL INCREASE TO 5O CENTS ON THE FOURTH 
 DAY AND TO $1.OO ON THE SEVENTH DAY 
 OVERDUE. 
 
 BEG 12 1933 
 
-4- 
 
 UJ99