GIFT OF - ;nter of illiam Stuart Smith \t*s~4* THE INCANDESCENT LAMP AND ITS MANUFACTURE. BT GILBERT S. RAM, A.I.E.E. NEW YORK : THE D. VAN NOSTRAND COMPANY, 23, MURRAY STREET, AND 27. WARREN STREET. LONDON : THE ELECTRICIAN" PRINTING AND PUBLISHING COMPANY, LIMITED, SALISBURY COURT, FLEET STREET, E.C. 1894. [All Rights Reserved.] \ Printed and Published by THR RLRCTRIOIAN" PRINTING AND PUBLISHING CO., LIMITED, ], 2, and 3, Salisbury .Court. Fleet Street, London, E.C. CONTENTS. PACK LIST OF ILLUSTRATIONS vii. INTRODUCTORY ix. CHAPTER I. THE FILAMENT 1 5 Carbon Filaments. Filaments of other Substances than Carbon. Volatilisation of Carbon. Disintegration of Carbon. Emissivity of Carbon. Methods of Preparing Carbon. CHAPTER II. PREPARATION OF THE FILAMENT 6 : 19 Swan's Process of Parchmentising Cotton Thread. Drying Prepared Thread. Draw Plates for Cutting Thread. Squirted Filaments. Wynne and Powell's Process. Weston's Process. Furfurol Process. Cuprammonia Process. Other Processes. CHAPTER III. CARBONISATION 21 32 Precautions necessary on account of Shrinkage. Filaments Carbonised on Blocks. Double-loop Filaments. Method of Packing Filaments in Crucible. Flat Filaments. Construction of Furnace. Pyrometer. Precautions during Carbonisation. Unpacking the Crucibles. CHAPTER IV. MOUNTING 33 44 Use of Platinum. Various Methods of joining the Filament to the Wires. -Mechanical Joints. Deposited Carbon Joints. Socket Joints. Butt Joints. Apparatus and Details for making Deposited Joints. A 870733 IV. CONTENTS. CHAPTER V. PAGE FLASHING 4567 Uses of Flashing. Flashing to Resistance (hot or cold). Effects of Flashing. Flashing in Gas at Atmospheric Pressure. Vacuum Flashing. Gasoline or Pentane Flashing. Quality of Deposited Carbon. E.M.F. required for Flashing. Extent of Surface with Flashed and other Carbon. Effect of Reduction in Resistance due to Flashing on the Voltage, &c. CHAPTER VI. SIZES OF FILAMENTS (UNFLASHED) 69 79 Circular Filaments. Square Filaments. Flat Filaments. Hollow Filaments. CHAPTER VII. SIZES OF FILAMENTS (FLASHED) 81 97 Curves showing Sizes Before and After Flashing. Filaments with Resistance Cold Double the Hot Resistance. CHAPTER VIII. MEASURING THE FILAMENTS 99 102 Screw Micrometer Gauge. Trotter Gauge. V-slot Gauge Optical Method. CHAPTER IX. GLASS MAKING 103 108 Glass Furnace. Crucibles. Making Lamp Bulbs and Tubing. CHAPTER X. GLASS BLOWING 109 1 1 9 Cannon Blow-Pipe. Large and Small Flames. Precautions in Working Lead Glass. Compound Blow-Pipe Air Supply for Blow-Pipes. Lamp Bulbs Blown from Tubing. CHAPTER XI. SEALING-IN 121129 Description of Process. Annealing. Spotted Bulbs. Cutting Glass-Tubing. Grinding Stoppers, &c. CONTENTS. V. CHAPTER XII. PAGE EXHAUSTING 131 177 Necessity for Exhausting. Mercurial Air-Pump. Pumps of Geissler, Laue-Fox, Toepler, Swinburne, Sprengel. Pumps Arranged for Factory Work. Methods of Lifting the Mercury. Shortened Pumps. Weston's Pump. Sawyer- Mann Pump Room. Kennedy's Pump. Application of Current to Lamps during Exhaustion. Heating the Lamps. Air Film in Pumps. Method of Constantly Maintaining Vacuum in Pump. Moisture in Lamps and in Pumps. Method of Joining the Lamps to the Pumps. Pump-Room Blow-Pipe. Testing the Vacuum. McLeod Vacuum Guage. Induction Coil Test. Clearing and Distilling Mercury. Mechanical Pumps. Giminghain's Pump. Rotary Pump. CHAPTER XIII. TESTING 179 192 Testing by Eye. Photometer Testing. The Photometer. Wattmeter. Meaning of Candle-power. Candle-power of Flat Filaments. CHAPTER XIV. CAPPING 1 92 194 Caps. Methods of attaching by Plaster. CHAPTER XV. EFFICIENCY AND DURATION 195 201 Temperature and Efficiency. Falling-off in Candle-power. Effect of Variation in Pressure on the Candle-power. Causes of Falling-off in Candle-power. Effect of Fall in Resistance. CHAPTER XVI. RELATION BETWEEN LIGHT AND POWER 203 210 Tests of Four Lamps. Calculation of Candle power at any Voltage. INDEX TO CONTENTS 213 LIST OF ILLUSTRATIONS. FIG. PAGE 1. Apparatus for Parchtneutising Cotfcou Thread (Swan's Process) 8 2. Glass Guide Rod for Thread ... 10 3. Method of Drying Parchmentised Thread .. 13 4. Circular Jewelled Draw-Plate for Cutting Parchmentised Thread 15 5. Split Jewelled Draw-Plate for Cutting Parchmeutised Thread 15 6. Carbon Frame for Carbonisation of Single-Looped Filaments .. 22 7. Carbon Frame for Carbonisation of Double- Looped Filaments 23 8. Carbonising Furnace ....... ... . ... .. 26 9. Pyrometer for Indicating Temperature during Early Stages of Carbonisation .. 27 10. Early Swan Lamp, showing Crayon-holder Joint .. ... 35 11. Early Lane- Fox Lamp, showing Carbon Tube Joint 36 12. Early Maxim- Weston Lamp, showing Bolt and Nut Joint 36 13. Wire-Twisting Machine 37 14. Cementing Machine for Socket Joints ... ... .. 39 15. Cementing Machine for Butt- Joints . ..:' ... 40 16. Apparatus for Coal-Gas Flashing 52 17. Apparatus for Pentane Flashing ... ... ... ... 55 18. Apparatus for Pentane Flashing .. 55 19. Clip for Holding Filament during Flashing 56 20. Vacuum Connections for Pentane Flashing ... ... ... 57 21. Curves showing Effect of Flashing on the Resistances of Filaments of Various Diameters ... 65 22. Curves showing Effect of Reduction in Resistance, due to Flashing, on the Voltage of Filaments of Various Diameters 65 23. Curves showing Effect of Reduction in Resistance, due to Flashing, on the Current for Filaments of Various Diameters 66 24. Curves showing Relation between Current and Diameter of Filaments for Different Durations of the Flashing Process. The Diameters represent the Measurement of the Fila- ments before being Flashed 82 25. Curves showing Relation between Current and Diameter of Filaments for Different Durations of the Flashing Process. The Diameters represent the Measurement of the Fila- ments after being Flashed ... ... ... ... ... 83 26. Micrometer Gauge for Measuring Diameter of Filaments ... 99 27. Trotter's Micrometer Gauge .. ... ... ... ... 100 28. V -slot Micrometer (Jauj^e.. 101 29. Glass Crucible for ' Cone " Furnace 104 viii. LIST OF ILLUSTRATIONS. FIO. PAGE 30. Glass Crucible for Small Furnace 105 31. Bulb Mould 107 32. Cannon Blow-Pipe 110 33. Form of Large Flame of Cannon Blow-Pipe Ill 34. Pointed Blow-Pipe Flame ... Ill 35. Compound Blow-Pipe ... ... ... ... ... ... 113 36. Form of Flame of Compound Blow-Pipe ..' ... ... 114 37. Different Stages in Blowing a Bulb from Glass Tube ... .. 117 38. Different Stages in the Process of "Sealing-in " the Filaments 122 39. Lamp with Annealing Cap "... .. ... .. 124 40. Geissler Pump . 134 41. Lane-Fox Pump ... ... ... 137 42. Toepler Pump ... 138 43. Safety Side Tube Arrangement for Use with Geissler Class of Pump .'. . f . ... 141 44. Swinburne Pump ... ... . ... .. ... ... 142 45. Sprengel Pump. Simple Form ... ... .. ... 144 46. Sprengel Pump Distributor for Six Fall Tubes 145 47. Sprengel Pump Collector for Six Fall Tubes 146 48. Sprengel Pump with Air- Pressure Lift i49 49. Shortened Form of Sprengel Pump ... 150 50. Weston Pump .. 153 51. Pump Room of Sawyer-Mann Lamp Factory, 1889 ... ... 154 52. Swinburne Pump. Factory pattern .. 155 53. View of a Pump-Room with Swinburne Style of Pump ... 156 54. Details of Top of Mercury Reservoir of Pumps in Fig. 53 .. 157 55. Kennedy's Pump ... ... .. ... ... . ... .. 159 56. Swinburne's Arrangement for Constantly Maintaining a Vacuum in a Pump ... ... ... .. ... ... 163 57. Blow-Pipe for Pump Room ... ., ... 169 58. McLeod Vacuum Gauge ... ... ... .. . . 170 59. Mercury Still ... 175 60. Electrical Connections for Testing Lamps on Photometer ... 187 61. Evershed's Wattmeter for Lamp Testing. Sectional elevation 188 62. Evershed's Wattmeter for Lamp Testing. Sectional plan ... 188 63. Connections of Evershed's Wattmeter ... . .. 189 64. Curve showing Percentage of Normal Candle-Power at any given Percentage of the Normal Volts 198 65. Curves showing Relation of Candle-Power and Watts for Four Lamps, A, B, C, D 205 66. Curves showing Relation of Candle-Power and Watts for Lamps B and D, up to their Breaking Points ... .. 205 67. Diagram showing Relation of the Logarithms of the Candle - Powers to the Logarithms of the Volts and of the Watts respectively of Lamps B and D ... ... ... ... 207 68. Lower part of Curves of Candle-Power and Volts for Lamps BandD 208 69. Curves of Candle-Power and Volts for Lamps B and D up to their Breaking Points ... 209 INTRODUCTORY. WITH the expiration of Edison's master-patent for the carbon incandescent electric lamp, the attention of electric light engineers, as well as of all those who use the light, is once more directed to the consideration of the lamp itself, to the possibility of obtaining better lamps, and to the probable reduction in price which will naturally follow. Owing to the long prevailing monopoly in the sale and manufacture, there has been little inducement for those interested to ex- periment and to study the problems connected with the incan- descent lamp. As a result of this, the literature of the lamp is very scanty, and is entirely confined to the pages of the leading technical journals. While dynamos, alternators, transformers, arc lamps, and almost every piece of apparatus connected with electrical engineering and lighting, have been written on at length and discussed at meetings of scientific societies, the incandescent electric lamp, which has been the chief cause of the very existence of these machines and apparatus, has been comparatively neglected. With the exception of the valuable series of articles by Mr. Swinburne which appeared in The Electrician six years ago, no comprehensive or detailed account of lamp manufacture has appeared. The manufac ture of the incandescent lamp and the principles underlying it are, consequently, but little known, except to those actually engaged in the work. X. INTRODUCTORY. As a thorough understanding of the lamp and the possi- bilities of its improvement can only be obtained by consider- ing the various processes of its manufacture, it is probable that a work on the subject at the present time will be welcomed by those who are interested therein and have not had the opportunity of studying it for themselves. As writers in whose hands this most interesting subject might have fared better have not essayed to undertake the task, the Author asks the indulgence of readers for the many shortcomings which may be apparent to them. This indulgence will, he feels, be the more readily extended to him when those interested in the subject understand that " The Incandescent Lamp and Its Manufacture " does not profess to at all exhaust the subject, or to describe nearly all the processes of manufacture. All that is attempted is to give readers such information as the Author, in the course of a considerable experience in lamp-making, has acquired, and to place this information before them with as little mathematical embellish- ment as, under the circumstances, is possible. CHAPTER I. THE FILAMENT. THE light-emitting portion or burner called by Edison the filament of the incandescent lamp, it is well known, is composed of carbon, which is raised to a state of incan- descence by the passage through it of a current of electricity. Carbon is not the only substance which can be used. The number of possible materials is, however, very small, and for two reasons. Firstly, because the material must be capable of withstanding a very high temperature; and secondly, because it must be a conductor of electricity. These two conditions exclude nearly everything except carbon and some of the metals. Of the metals which are sufficiently plentiful, platinum alone possesses the desired qualifications to a degree which renders its use possible. Platinum has a very high melting point, and will stand a very high temperature, but the temperature necessary for an economical incandescent lamp burner is far above that even of the melting point of that metal. The incandescent electric lamp, though a great advance upon most other artificial sources of illumination, gives us at best only a very inefficient method of obtaining light. That is to say, the proportion of the luminous radiation to the total radiation is very small. When the temperature of a filament is raised by the expenditure in it of more and more energy it becomes increasingly efficient. The efficiency which can be attained is, however, limited by the temperature which the filament will stand. Carbon will stand a temperature higher than the melting point of platinum, and it, therefore, con- stitutes a more efficient burner. 2 THE INCANDESCENT LAMP AND ITS MANUFACTURE. Certain metallic oxides possess in a remarkable degree the power of resisting very high temperatures. It has, in conse- quence, been proposed to utilise these in some way, and mention [<$ them jiais^eeii made in some of the earliest patents for elec- tric^ lanrjss. Here, .however, the difficulty occurs that metallic oxides arfc daq^-js^ndjq^tors of electricity, and, consequently, cannot be used alone. They can only be used in conjunction with some conductor either a metal or carbon. The Author has tried coating both platinum and carbon with oxides such as alumina, magnesia, lime, &c., but has always met with the difficulty, in the first place, that the coefficient of expansion of the oxide is different from that of the carbon or metal, and, consequently, the coating cracks and falls off. It has been pointed out by Prof. S. P. Thompson that the temperature of the crater of the arc in an arc lamp is con- stant, and is the temperature of the volatilisation of carbon. Any other substance contained in the carbons volatilises instantly when reaching the arc. The carbon itself volatilises at that temperature, and all other substances appear to do the same either at that temperature or below. This being the case, it would appear futile to attempt to increase the durability of arc lamp carbons by the addition of any foreign substance. It is possible, however, that some substance other than carbon may yet be used successfully as the light-emitter in an electric lamp. Even if all metallic oxides are dis- persed at a temperature less than that of the arc when used in conjunction with carbon, they are not all necessarily volatilised at that temperature. The volatilisation may be simply a secondary effect, the oxide having been first reduced by the carbon. It does not appear to be impossible that some oxide or mineral substance may be found to be stable at tem- peratures higher even than that of the arc if there be no carbon present. The great trouble in using any such sub- stance in the same manner as the filament of an ordinary incandescent lamp is that of electrical conductivity. It is possible that some of them might consent to conduct suffi- ciently if subjected to a very high E.M.F., such as can be produced by transformers. We have seen quite recently slate pencils used in the place of carbons in the arc THE FILAMENT. 3 slate being considered an insulator at ordinary electrical pressures. An incandescent lamp, however, with such a filament, if ever constructed, would hardly be suitable for domestic use. Again, although mineral substances when mixed with carbon are all reduced or volatilised in the arc, such mixtures may, nevertheless, be used for incandescent lamp filaments at a temperature much below that of the arc. Great improvements may yet, however, be looked for in the carbon itself. It is well known that carbon prepared by some processes will last much longer than that prepared by others. The efficiency of a carbon depends upon the tem- perature at which it can be run. The higher the temperature the greater the efficiency. The limit to the temperature possible is that of the volatilisation of carbon. This tempera- ture in the case of an incandescent lamp is probably lower than that of the arc. The temperature of volatilisation of carbon in vacuo is likely to be much less than at atmospheric pressure. We can, therefore, never hope to get the carbon of an incandescent lamp anything like as bright as the crater of an arc lamp. Long before the temperature of volatilisation of carbon in vacuo is attained, there is another effect to be considered which puts a far lower limit upon the temperature which can be maintained in practice. Apart from volatilisation properly so called, there is an action going on by which particles of carbon are dissociated from the filament and thrown off upon the glass globe (which is usually called the bulb) of the lamp. This action, unfortunately, begins at a comparatively very low temperature a little above red heat and increases very much at higher temperatures, so that, although the filament may be far below the temperature of volatilisation, yet it is fast falling to pieces, and the bulb is becoming greatly obscured by the black deposit of these particles of carbon. It is, then, in overcoming this lower limit to the temperature that improvements must first be made. Some carbons are much better than others in this respect. The best carbon is that which at the highest possible temperature disintegrates at the slowest rate. Much has been said and written about the so-called efficiency of surface, or emissivity, of different varieties of B 2 4 THE INCANDESCENT LAMP AND ITS MANUFACTURE. carbon. Some varieties have a greater emissivity than others in the proportion of about 3 to 2. That is to say, when running at the same efficiency (watts per candle), two carbons of precisely the same dimensions may be giving different amounts of light by as much as the above ratio. Or two carbons, one having only two-thirds of the surface of the other, may, when at the same efficiency, be giving an equal amount of light. It has sometimes been supposed that the one with the smaller surface is, therefore, better than the other. The reverse, however, is usually the case. The fila- ment with the smallest emissivity is generally the best. A high emissivity usually means a dead black surface, a low emissivity a polished white one. The dull black surface is generally a softer carbon, and one which disintegrates more quickly than the other. The only advantage in the high emissivity is that the filament is, therefore, shorter and may be accommodated in a smaller bulb. The size of the bulb, however, should not be determined by the length of the filament so much as by the candle-power. A large candle- power lamp may have a short filament, but it should not for that reason be put into a small bulb. These matters will be further dealt with in a subsequent chapter. The chief aim of the lamp-maker, next to that of producing a durable carbon, is to make filaments which will be all alike when finished. To be commercially successful a lamp-maker must be able to make his filaments within a very little all alike in every particular, that is in dimensions (length, circum- ference, cross-section), in quality of surface, and in specific resistance. Many different methods of preparing carbon have been proposed. Whatever the method, there is one operation which almost all of them necessarily include that of roasting or baking the material in a furnace out of contact with the atmosphere, in order to drive off, as far as possible, every- thing contained in the material which is not carbon. In most cases an entire change in the nature of the material is produced by the process. One of the first methods to be tried was that of forming the filament out of a paste made of ground carbon, with a binding agent such as a solution of sugar or caromel. Such a paste was squirted or moulded to the desired shape, dried, and then baked. This, of course, THE FILAMENT. 5 produced practically the same material as is used for the carbons of arc lamps. The difficulty in this method is to produce fine and uniform filaments. In any event, however, the resulting carbon is not compact enough for the purpose, and it rapidly disintegrates when used in a lamp. Edison proposed a mixture of lamp-black and tar, an utterly worth- less process in itself, but one which has, nevertheless, become famous in the history of lamp-making. Processes such as these involve the baking of the material in order to drive off the volatile matter contained in the binding agent. The successful processes are not those by which the fila- ments are formed direct from a substance containing carbon as such, but from a material containing carbon in chemical combination, such as silk, hair, woody fibre, or pure cellulose. Such substances require baking at a high temperature in order to reduce them to carbon, their other constituents being driven off by the heat. The process of baking in this case is generally called carbonisation. The only processes which do not include that of baking or carbonisation are those by which the filaments are deposited or built up from the carbon contained in some vapour or liquid such as illuminating gas or petroleum oil. CHAPTER II. PREPARATION OF THE FILAMENT. SOME of the methods of preparing filaments will now be described. In spite of all the work that has been done by experimenters during the last ten or fifteen years with a view to providing something new and better, it remains a fact that the original process of Swan's, that of parchmentising cotton thread by sulphuric acid, remains to this day one of the best, if not the best and most easily worked of any. It will, there- fore, be described first. The raw material in Swan's process is pure cotton, such as is sold for knitting, loosely spun into a thread. It is usual, in the first instance, to clean the thread by boiling it hi soda or ammonia, in order to take out any grease which may be present in it, and which would prevent the sulphuric acid from acting on those parts which would be thereby protected. It must, however, be thoroughly washed afterwards, in order to remove all traces of the alkali, and be then dried. It is, after all, an open question if the above process of boiling in alkali does any good, provided that the cotton is as clean as it can be obtained in the first instance, for in spite of the process, there are often found to be places in the thread upon which the acid will not act, or rather upon which it will not act in the tune allowed. Sometimes these places are very small, and can only be seen by examining the parchmentised thread very closely, while at other times they will extend for several inches along the thread, and occur at short intervals. No satisfactory explanation of these acid-proof, or partially acid-proof, sections has yet been given, or, at any rate, no satisfactory remedy has been proposed. THE INCANDESCENT LAMP AND ITS MANUFACTURE. PREPARATION OF THE FILAMENT. V When dry, the cotton thread is wound on a drum, and is afterwards unwound and drawn through a dish of sul- phuric acid of the required strength, whence it is drawn into water and wound on another drum, which is preferably wholly under the water. This sounds a very simple and easy process, and is, provided that certain precautions are taken. In the first place, the thread, after passing through the acid, is extremely weak, and must, accordingly, be drawn along as regularly and with as little friction as possible. To insure this, the drum upon which the thread is wound after passing through the acid, and by which it is drawn along, must be driven by power, and must be so geared as to run very steadily and regularly without any jerking. The slightest friction any- where will put a strain on the thread and will cause it to break. Fig. 1 shows an arrangement which the Author has found to work very well. At the left-hand end is seen the drum A, upon which the thread is wound in the first instance. K is the drum on which the thread is ultimately wound and which draws it along. The thread has to be guided in its passage from one drum to the other, so as to pass for the right length of time through the acid and then into the water. It must be remembered that the action of the acid continues from the moment when the thread enters the acid until it enters the water, though the thread is only actually under the acid in the dish for a short time, after which it passes through the air until it reaches the water. In order to guide the thread with the minimum of friction, glass rods with an open loop formed at the end (Fig. 2), may be used, the thread passing through the loop. The advantage of this form is that the thread may be put into the loop while in motion without having to thread the end through first. A number of blocks (D, E, F, G, Fig. 1) are mounted on a bar about 18in. above the bench, each capable of sliding lengthwise along the bar. To each block one of the glass guide-rods is attached, so that it may slide up or down and be held at the required height by a clamp. Between the left- hand drum A and the first guide-rod D, a sheet of glass, B, is laid on the bench. H is a dish containing the acid. J is a 10 THE INCANDESCENT LAMP AND ITS MANUFACTURE. lead-lined trough of some 6ft. in length, terminating in a tank L, which contains the winding drum K, driven from overhead by a band. In starting the apparatus, sufficient thread is wound off the drum A, so that the end will reach to K. More thread is then wound off A, and laid zig-zag fashion across the glass- plate B. The first length of thread is then dropped into the glass guide-rods and the end is looped on to a projection on the edge of the drum K as it revolves. The thread is, there- fore, wound on to K, and drawn along from off the plate B. One of the guide-rods, say F, is now lowered so that it depresses the thread below the surface of the acid in the dish H. On emerging from the acid, the thread passes up through the guide G, and then drops by its own weight into the water in the trough J. The object of the trough J is FIG. 2. Glass Guide Rod for Thread. that the thread may be examined easily in order to see if it is receiving the proper amount of treatment. It can be best examined while passing along this trough, as it is readily seen in the water against the lead lining of the trough. On coming out of the acid dish, the thread has a transparent and bright appearance, like a string of jelly. When it enters the water it is immediately changed from the transparent to a semi-opaque condition. It is at this point, after entering the water, that it must be constantly watched to see if it is receiving the proper amount of treatment. If it is not being- done sufficiently, it will have a core of unconverted cotton running through it continuously or in patches. If it is receiving the proper amount of treatment it will appear as a semi-opaque homogeneous thread. On the other hand, if PREPARATION OF THE FILAMENT. 11 the treatment is too long the thread will probably break. It is, therefore, necessary at starting to give it less than the proper amount of treatment. The acid dish can then be moved gradually further back, so that the thread has a longer distance to travel from the acid to the water. If more treat- ment than can be given by this means is required, the speed of the drum K must be reduced. When the apparatus is adjusted, the acid dish, the edge of which is ground level, is covered by pieces of sheet-glass so that the acid is not exposed to the air, only a small opening being left just where the thread enters and leaves. As the thread is drawn along, it is, of course, taken up off the glass plate, and the attendant must take care to have more thread wound off the drum and laid across the glass plate before the last lot is used up, or a breakage will occur. Another plan is to drive the drum A as well as K by power. The thread is then unwound from A as it is wanted. The former method, however, gives less trouble. By . the method of laying out the thread in zigzag rows on the glass plate the attendant has sufficient time to examine the thread as it passes along the trough and to keep the apparatus working properly. A sheet of glass is required, because the fine fibres of the loosely- spun thread are apt to < carbon and wires, and must consequently be driven off again before the filament is sealed into the bulb. The first thing to be done towards making the joint is to prepare the platinum wires. The wire has first of all to be cut up into the lengths required. A deposited or paste joint can then be made without any further work on the platinum, but it is better to prepare the platinum in some way, or the cement will not have a good hold upon the wire. The plain wire is too smooth, and should be roughened or shaped so that the carbon cement has something to hold on by. The wire may be twisted into a spiral, forming a socket into which the end of the filament is put. The spiral can be formed by twisting the wire with a machine like that shown in Fig. 13. FIG. 13. Wire-Twisting Machine. A A is a spindle, with a needle B at one end and a handle C at the other. There is a screw thread on A which allows the needle to be rotated the same number of times (two or three usually) for each wire. The end of the platinum wire is in- serted in a small hole close to the needle. The wire is held tightly, and the spindle rotated by the handle, thus twisting the wire two or three times round the needle. The wire is then slipped off the needle, straightened, and finished off with a small pair of pliers. If a socket is required, it is better to make the tube form, which uses less platinum than the spiral. To make the tube, the wires are first flattened out for an eighth of an inch or so at the ends. This can be done by means of rollers, which may be arranged to cut the wires at the same time. The wire, being fed into the machine in a continuous length, is cut 38 THE INCANDESCENT LAMP AND ITS MANUFACTURE. and flattened at one operation. The wires are then drawn through an ordinary wire draw-plate, which bends the flattened ends round into a tube. Before drawing through the plate, the flattened ends must be annealed by heating to redness in a Bunsen gas flame. If this is not done these ends will be pulled off instead of being formed into a tube. The filament is slipped into the socket, which is then squeezed upon it so as to hold it tightly. It is convenient to have the wires joined together by a bridge of glass before mounting in the socket, but it is not necessary. If a butt or lap joint is to be made, the wires may be nicked near the end, so as to form an indentation by which the deposit can hold on. An excellent way of making a butt joint is that which is adopted by the Edison- Swan Company. A small, flat head, like a pin-head, is formed on the end of the wires. The end of the filament is placed against the centre of this head, and carbon is deposited over the head and the end of the filament. As the carbon is deposited so as to completely cover up the head, it forms a very secure joint. Deposited carbon will always hold tightly to the filament. For filaments taking up to an ampere of current, platinum wire of 0'014m. diameter is generally used. The Author has, however, known lamps taking several amperes to work perfectly well with this size of wire. For currents above one ampere larger wire should be used, and for more than five amperes several small wires are better than one large one. The deposited carbon joint, as has already been mentioned, is made by strongly heating parts of the wires and the fila- ments by a current of electricity in a hydrocarbon vapour or liquid. The hydrocarbon is decomposed by the heated wire and filament, and a deposit of carbon upon the part heated is the result. In order to make such a joint it is necessary to fix the wires and filament in a machine which will make the necessary electrical connections. For filaments which are mounted in sockets the illustration Fig. 14 shows a convenient apparatus. The filaments being already clamped in the sockets on the wires, the wires are inserted under the springs A A, which can be raised by the pins B B (one only of these two springs with its pin can be seen in the illustration). The metal pieces CC, which turn stiffly about DD, are turned so that MOUNTING. 39 the parts E E lightly press on the filament at a short distance beyond the platinum wire. The spring contact pieces F F are then lowered by the thumbscrews G G, so that they press upon the filament just over the places where it is supported by EE. A good contact is thus secured upon the filament without straining it. The body of the instrument, K, may be made of wood, and all the metal pieces may be of brass. Between the springs A A and the wood K are metal strips L L, which are fastened to K and are continued beyond K for some FIG. 14. Cementing Machine for Socket Joints. distance. The current enters and leaves the apparatus by these pieces, which may conveniently terminate in a hook H. The path of the current is from the hook H along L and A to the platinum wire, along the filament to F, round F to the other side of the filament and to the other platinum wire, and back through the corresponding parts on the other side of the instrument to the second hook H. The larger part of the filament is thus short circuited by FF, and the process is consequently sometimes called "short-circuiting." If a butt or a lap joint is to be made, the machine must be arranged so that the filament can be easily and quickly 40 THE INCANDESCENT LAMP AND ITS MANUFACTURE. adjusted to its position opposite the platinum wires. For this purpose an instrument like that shown in Fig. 15 will answer the purpose. There are two sets of spring contacts A A and B B, mounted on a wood block C, of which that part between the two sets of springs is cut away to the depth of about half an inch. The electrical connections are made to the spring con- tacts A A by means of the wire D and a corresponding wire on the other side of the instrument. The platinum wires are first clamped in A A, and the filament is then put into B*B, so that its ends touch the ends of the platinum wires. The part which gets heated by the current is, of course, that be- tween the two sets of clamps. The block is cut away to allow of a free circulation of the hydrocarbon. FIG. 15. Cementing Machine for Butt Joints. When the deposit is obtained from a hydrocarbon vapour, the apparatus, either that of Fig. 14 or Fig. 15, with the filament and wires already placed in position, is hung on a support, which also conveys the current. This support is fixed on a flat circular stand, which has a groove running round it. This groove contains mercury, and forms an air-tight joint for the glass shade, which is placed over the apparatus. Common coal gas, enriched by passing over benzoline, gasoline, or, ether, is led into the apparatus through the stand, and an exhaust pipe is also provided. The gas is turned on, and when sufficient time has been allowed for it to fill the enclosed chamber the electric current is turned on. A resistance is, ; used, so that the current can be gradually increased as the deposit proceeds. The glass shade must be weighted, or other- MOUNTING. 41 wise held down, or it may be lifted up by the gas inside. The current is regulated so that the pieces of the filament between the contacts and the platinum wires are brought to a bright red heat, and the deposit commences. It takes, however, a long time in this way to get a sufficient deposit upon the platinum, and it is, therefore, better to use a little carbon paste on the junction. A suitable paste may be made by mixing powdered carbon with a solution of dextrine or caromel, and can be applied with a small brush. If this is done the current will be increased much more quickly, and a sufficiently strong joint be obtained. The liquid process is, however, more easily worked, and makes a more satisfactory joint. In this process the frame which holds the filament and platinum wires is hung on a support which also conveys the current, so that the part which is heated, and upon which the deposit is to be formed, is entirely below the surface of the liquid. Several fila- ments may be cemented at the same time, the frames being arranged in series, with a switch connected to each, so that any one of them may be cut out if necessary. It is more con- venient to use a separate jar of liquid for each filament, though one large trough will answer the purpose. The jars are set in a metal or stoneware trough to catch any spillings of the liquid. It is also advisable that the whole of the bench at which the work is done be covered with zinc to prevent the wood from being soaked with the oil. An extinguisher, to cover over the whole of the trough containing the pots of liquid, should be provided in case of accident. There is, however, no danger from fire, unless very inflammable liquids are used. Any hydro- carbon liquid will produce a deposit on a carbon filament heated to redness below the surface of the liquid, though some liquids deposit very much more quickly than others. Vegetable oils, such as olive oil and linseed oil, produce a good deposit. They are, however, too viscous for use with very fine filaments, and the smoke which is produced has a bad smell. Turpentine deposits very rapidly, but the deposit is too soft. Benzoline or benzine rapidly deposit, but are too dangerous for ordinary use. Ether gives a rapid deposit, but, of course, cannot be used in the open. Mineral oils are generally used. The best kerosene oil gives a good and hard deposit, but is very slow. 42 THE INCANDESCENT LAMP AND ITS MANUFACTURE. Crude petroleum, a black liquid containing mineral oils of various densities and flashing points, answers the purpose very well. A satisfactory liquid may be obtained by mixing one which deposits rapidly with one which does so slowly, but gives :a hard deposit ; such a mixture, for instance, as four parts of best (high flashing point) kerosine to one part of turpentine. This mixture gives a sufficiently rapid and hard deposit, and is not dangerous to use. Even when the liquid has become very hot, a lighted match applied to the surface will be extinguished. While the deposit is going on a stream of dark smoke rises from the liquid. This smoke will burn as it bubbles up out of the liquid if it be lighted, but it can easily be blown out again. To obtain a proper deposit the joint is maintained at a bright red heat. There is no danger of firing the liquid so long as the joint is kept below the surface, and the current turned off before it is lifted out. The liquid may be fired by carelessness on the part of the operator. If, for instance, a joint breaks, as sometimes happens through excess of current when the depositing is going on, and so breaks the circuit, the operator may take out the filament and put in another, having forgotten to turn the switch off. As soon as the form with the new filament touches the contacts the joint will light up, and if it is not below the surface of the liquid, it will certainly set fire to the oil, if it be an inflammable one. The strength of current used in making a joint is many times greater than the joint will ever be called upon to carry afterwards. In order to join a filament six mils in diameter to platinum wires of fourteen mils with tube sockets, the deposit to extend for a tenth of an inch along each leg of the filament, about twenty volts will be required to start the depo- sition, i.e., to make the ends of the filament bright red-hot under the oil. As the deposition of carbon proceeds, the ends of the filament become rapidly duller. The thickness of the filament is increased, and, therefore, the cooling surface, while the resistance is at the same time diminished; and as there is a regulating resistance in the circuit, having a considerable resistance compared with that of the joint, the current rises only a very little. It is, therefore, necessary to cut out some of the regulating resistance and increase the current, so as to keep the joint at a bright red heat. In this way the current MOUNTING. 43 will be gradually raised to at least twelve amperes before the joint is large enough, with a sufficient amount of deposit on the platinum. The volts will fall as the current is raised. With the best kerosene oil it will take ten minutes to make this joint. If the operation be hurried by turning up the current too fast, the joint will probably break, and both the filament and the wire will be spoiled by the arc which will be formed at the break. With one-fifth part of turpentine added to the kerosene, the time occupied can safely be reduced to one minute and yet produce a good hard joint. A circulation of air should be arranged in the room where the process is carried on, in order to carry the smoke and fumes away from the operatives. The fumes from mineral oils do not appear to be unwholesome as are those from animal or vegetable oils, although the smell is not altogether pleasant. The resistance used for the purpose of varying the current may be composed of a number of carbon plates lying one against another, with a screw arrangement for tightening them ; or a water resistance may be used. The least trouble- some way, however, is to use an alternating current with choking coils instead of resistance. An ammeter should be used, so that the operatives can watch the rate of increase of the current and cut it off at the right point. When the liquid is fresh, the current may be regulated according to the ap- pearance of the red-hot joint, but when the liquid has been used for some time and becomes black, the hot joint cannot be seen. Even then, however, the regulation can be effected by watching the volume of the smoke which bubbles up. An ammeter is, however, more reliable. The deposit formed on the platinum is due to the plati- num being heated to the required extent more by conduction from the hot ends of the filament than by direct heating by the current. The deposit, consequently, does not reach far along the platinum, which is kept too cool by the liquid to decompose it. After this process the filaments and platinum wires will be oily and must be cleaned. This is easily done, as regards the filament, by gently heating by means of current, while the platinum wires can be heated in a Bunsen or blow-pipe flame. 44 THE INCANDESCENT LAMP AND ITS MANUFACTURE. If a thick, sticky oil has been used for depositing' from, it may be necessary to first wash the filaments in kerosine. In making a paste joint, a drop of the carbon paste is applied to the ends of the platinum wires with a small brush, and the filament is stuck into the drop, which quickly dries and sets. The paste must then be heated to drive off the volatile substances with which the carbon powder is mixed, and the joint is made. CHAPTER V. FLASHING. THE process usually called " Flashing" will now be con- sidered. It is a similar process to the one just described of making joints by the deposition of carbon. Carbon is deposited upon the whole of the filaments from either a liquid or gaseous hydrocarbon. The name of ''flashing" was given to the process because the filaments used to be flashed, i.e., lighted up and extinguished a number of times at short intervals, while in a hydrocarbon vapour. The actual flashing in this way is not necessary to the process except under certain con- ditions, and the term will here be used to mean the depositing of carbon on a filament by electrically heating it in a hydro- carbon liquid or vapour. In the early days of lamp making, flashing was resorted to for the purpose of making the carbon filaments light up evenly along their length. Early filaments were not even, and would light up with bright and dull spots all over them. The process of flashing had the effect of reducing this unevenness, and, if carried on for a sufficient length of time, of entirely removing it. The reason is simple. The bright spots on the filament are mainly due to those parts having a higher electrical resistance than the rest, and the dull spots to a lower resistance. This variation in the resistance at different parts of the filament may be due to a variation in the specific resistance of the fila- ment, or to a variation in the thickness, or a combination of both these causes. Now, carbon is not deposited from a hydrocarbon until a certain temperature is reached, at which the hydrocarbon is decomposed, and, up to a certain point at any rate, the higher 46 THE INCANDESCENT LAMP AND ITS MANUFACTURE. the temperature the more rapidly does the deposition take place. Thus, when a spotty filament is lighted up in the presence of a hydrocarbon, the higher resistance parts, or bright spots, will be deposited upon before and more quickly than the rest of the filament. The result is that the resistance of those parts is reduced, and consequently they are less and less heated by the current, and are gradually reduced to the same brilliancy as the main part of the filament. As the current is increased so that the main part of the filament is hot enough to be deposited upon, its resistance will fall, and it will in turn be gradually brought down to the level of the dull spots. Filaments are, however, now made by several processes which will light up perfectly evenly along their whole length, and, therefore, do not require flashing for the purpose of obliterating spots. Another difficulty with early lamp-makers was to get all their filaments of the same resistance. The resistance of the carbon filaments would vary very considerably. Here, then, another use of the flashing process presented itself for reducing different filaments all to the same resistance. This can be done by connecting the filament during flashing with an ohm- meter, which will show the actual value of the resistance at any moment. When the resistance has fallen to the figure required, the current is cut off and the process stopped. Another method is to measure the resistance cold. A three-way switch is used to connect the filament alternately to the dynamo and to a Wheatstone bridge, by which the resistance is measured cold, using, perhaps, only one or two volts. The " bridge" is set for the required resistance, and a key is depressed. A reflecting galvanometer is conveniently arranged to throw a bright spot of light in front of the operator, who on depressing the key watches the behaviour of the spot. If the resistance is still too high after the first application of the current, the three-way switch is put over again, and the filament is again lighted up and a further deposit is given to it, and so on until its resistance is brought down to the required figure. The actual flashing current may be arranged to work a Wheatstone bridge. The resistances are constructed to with- stand, without undue heating, the currents which they will FLASHING. 47 receive, the filament itself forming one arm of the " bridge." A galvanometer with a conspicuous pointer can then be used and fixed where easily seen by the operator, who cuts off the current at the instant when the pointer indicates that there is no current passing in the galvanometer circuit. Just as filaments are now produced free from spots without being flashed, so are they also produced all alike in resistance, so that flashing is not necessary for either of these two purposes. Why, then, do lamp-makers continue to use the process? The reason is that the carbon deposited by the flashing process under certain conditions is so much more durable than anything which can be produced by other methods, that it improves most carbons to give them even an exceedingly thin coating of deposited carbon. It makes them mechanically stronger, and prevents the disintegration of the filament from proceeding as rapidly as it otherwise would do. There are, however, other effects of flashing which have to- be reckoned with. One of these is that the filaments are thickened by the deposit and their surface is consequently increased. Another is that the deposited carbon may and usually does materially alter the emissivity of the surface ; that is to say, the rate at which it will radiate light and heat. Thus flashing will at the same time reduce the resistance of the filament, enlarge the surface, and alter its emissivity. Now, to be commercially successful, a lamp-maker must be able to turn out lamps which are very nearly all alike in candle-power and voltage, when run at the same temperature ; or, rather, when run at the same voltage, the lamps must be equally bright, small differences in candle-power not being of great consequence. The lamp-maker has to aim at turn- ing out the filaments so that, at the required voltage, whatever it may be, they are all equally bright and approxi- mately of the same candle-power. Two or three per cent, variation in the voltage of the lamps is the maximum which should be permissable. A greater variation than this will be at once apparent to the eye by a difference in the bright- ness of the lamps. This is a much more important point than the actual candle-power, i.e., quantity of light being given out by a lamp. A small difference in colour or bright- ness can be seen at once, whereas a difference of 10 per cent. 48 THE INCANDESCENT LAMP AND ITS MANUFACTURE. in the quantity of light given out by two lamps running at the same temperature does not attract attention ; in fact, it can only be detected by the aid of a photometer. Again, suppose we take two lamps running at different temperatures. Let one be running at two and a-half watts per candle-power, and be giving a light of ten candles, and the other at four watts per candle-power, and be giving fifteen candles. Nine people out of ten will tell you that the two and a-half watt lamp is giving more light than the fifteen candle-power one, simply because it is brighter. Small differences in temperature are readily seen, while small varia- tions in the quantity of light are not noticeable. Hence a lamp-maker cares more about getting his lamps to run at the same temperature than at the same candle-power. A uniform temperature for all his lamps is the goal at which a lamp- maker has to aim. Now, as lamps are run in parallel, he also has to make his lamps run at a uniform temperature at a certain voltage. These are the two fixed quantities. The temperature may be whatever he considers best for the particular filament which he makes, so long as it is the same for all, and the voltage must be the voltage of the particular circuit for which the lamps are intended. The candle-power or quantity of light given out may vary to any extent among different lamps in one room, but if they are all at the same temperature they will look all right. Some of the different effects produced by flashing will now be briefly considered. To simplify the case as much as possible, it will be supposed that the filaments, after carbonisation, are all exactly alike in dimensions and in specific resistance, and in the nature of their surface or emissivity. Carbon deposited on a filament from a hydrocarbon may be of a soft, sooty, or of a hard and dense kind. The latter is the only kind that is any use, and will, therefore, only be con- sidered. Such carbon has a very much lower specific resist- ance than that of the filaments themselves. Consequently, with filaments of small cross section, a very slight deposit will reduce the total resistance very considerably. It is, therefore, FLASHING. 49 of the utmost importance that the process be arrested instantly the required resistance is reached. It is well known that the resistance of carbon falls as the temperature rises the reverse of the behaviour of the metals. Roughly speaking, the resistance of an incandescent lamp at the temperature of the air is double the resistance it will be when running at the ordinary temperature. Suppose that the resistance during flashing is indicated by an ohmmeter, and the current be cut off when it has fallen to a given number of ohms ; or let the circuit be arranged Wheatstone-bridge fashion, with a gal- vanometer to indicate when the required degree of resistance has been reached. If now a filament be flashed at a moderate temperature until the required resistance is indicated, and then another filament be flashed at a brighter temperature to the same resistance, and then the two filaments be made up into lamps and tested, they will be found to be quite different. At the same volts one will be brighter than the other. Why is this ? Simply because the resistances were adjusted at different temperatures. If both are run at the same tempera- ture the resistances will be found to differ. It is, therefore, obvious that this method of flashing to a certain resistance is useless, unless the flashing is done at the same temperature in all cases. The value of the method, therefore, depends on the skill of the operator in adjusting the strength of current by means of a variable resistance, so that the filaments are all flashed at the same temperature. It is difficult to do this correctly, as the filament is at a dazzling white temperature, and the operator's eye soon becomes unreliable. Of course, a darkened glass or spectacles may be used, but it is a question if the results are any better. Another trouble is that the glass receiver in which the filament is flashed becomes coated with a dark deposit which becomes thicker and obstructs more light with each filament that is flashed, and it must, conse- quently, be cleaned very frequently. At first sight, it might appear that if the resistances of the filaments be measured cold instead of hot, the error due to different flashing temperatures would be overcome. Here, however, another difficulty occurs. Carbon deposited at one temperature is of a different quality from that deposited at another, one difference being that its resistance temperature 50 THE INCANDESCENT LAMP AND ITS MANUFACTURE. coefficient is not the same. Therefore, two filaments flashed at different temperatures, so that they measure the same resis- tance when cold, will be found to have different resistances when hot. In this case again, the results depend on the accuracy of the eye of the operator for reproducing the same temperature during the flashing of each filament. Apart from these difficulties in flashing to a certain resis- tance, either hot or cold, there are many other ways by which the flashing may affect the ultimate evenness of the lamps. The deposit sensibly increases the thickness of the filaments, and consequently its radiating surface, and the thickened filament will therefore require a greater amount of electrical energy to maintain it at the required temperature. If the conditions of the flashing are reproduced exactly for each filament the thickening will be the same for all, and can be allowed for. If, however, a filament has been flashed to a certain resistance but has become thickened more than the amount allowed for, it will be found, when made up into a lamp, that although its resistance may be exactly what was wanted, yet it is dull when run at the voltage for which it was intended. If it be run to the proper brilliancy it will be found to be of greater candle-power than was intended. Again, flashing almost invariably alters the emissivity of the filament, or its power for radiating heat and light. If the conditions are precisely reproduced in each case a definite emissivity can be calculated upon. If not, and there is a greater emissivity than was allowed for, the lamps will be dull at the voltage for which they were intended, and if the emissivity be less than that calculated upon they will be over bright. It will now be readily understood that extreme care is needed in the process of flashing if uniform results are to be obtained. The actual amount by which the voltage of a lamp will be affected by any of these errors will be dealt with further on. In the meantime, some of the processes of flashing will be described. The filaments may be flashed either before they are mounted or afterwards, or both before and after. Almost any hydro- carbon may be used, but for each one there will be certain conditions which will give the best results. Thus, one will FLASHING. 51 work best if the filament be made very bright indeed, and another may only require a moderate temperature. One may require to be at a considerable pressure, while another must be much rarefied. It was stated in the chapter on " Mounting," that a heavy deposit is produced more quickly under a liquid than in a gas. For this reason it is best to flash in a gas, at any rate in the case of small-current filaments, as a thick deposit is not required. As a matter of fact, the deposition of carbon always takes place from the gas, even when done under the surface of a liquid. The liquid in contact with the filament is intensely heated, and by the time the filament is red hot the liquid round it is boiling so violently that the vapour alone comes into contact with it. It is this vapour which is decomposed and produces the deposit on the filament. There is produced in this way a hydrocarbon vapour of great density surround- ing the filament, and hence the rapid deposition of carbon. In reducing the resistance of a filament by flashing, it is neces- sary that the process be not too rapid, or a slight error in stopping the current at the right moment will considerably overstep the mark. For thin and high-resistance filaments, such as for 100-volt, 16-c.p., or 8-c.p. lamps, the process must be made much slower than for filaments for lamps which will take several amperes of current. It is consequently easier to flash low-current filaments successfully in a gas than under a liquid, and the gas may further be considerably rarefied. Ordinary illuminating gas can be used for flashing in at atmospheric pressure. In this case the apparatus required is very simple, as shown in Fig. 16. A is a glass bell jar ground flat on the top and bottom, so that with the aid of a little grease it stands gas-tight on the brass base-plate B. Through this plate pass the tubes C and D, one being the gas supply and the other the exhaust pipe. E is a plate made of hard wood, which carries two spring clips, G G, for holding the filament H. On the upper side of this plate are two metal contact plates, FF, connected with G G, the whole of this arrangement being removable. A spring contact arrange- ment, K, makes the electrical connection with F F, and at the same tune holds the plate E firmly upon the top of the bell jar so that it is gas-tight. To work the apparatus, a filament E 2 52 THE INCANDESCENT LAMP AND ITS MANUFACTURE. is put into the clips which are then inserted in the jar, and the spring contacts are let down on the top. Gas is then turned on through C, and air and gas escape through D. In a short time there will be very little air left in the jar, and the current may be turned on and gradually increased by means of a regulating resistance, until the filament is brightly lighted up. The deposition of carbon now begins, and the resistance of the filament becomes lower, and it gets much FIG. 16. Apparatus for Coal Gas Flashing. brighter. This, perhaps, at first sight appears curious, as it will be remembered that when depositing on a joint the joint became dull. Whether the filament or joint becomes brighter or duller as the deposition proceeds simply depends on the amount of resistance in series with it. In starting a high- resistance filament it probably requires all or nearly all the available electrical pressure to light it up, and consequently the regulating resistance is mostly if not all cut out. As the resistance of the filament decreases the current increases, while FLASHING. 53 the volts remain the same, or nearly so. The filament conse- quently becomes brighter, and the temperature must be kept down by adding resistance. On the other hand, if, when the filament is lighted up, there is a resistance in circuit greater than the resistance of the filament, it will become duller as the deposition proceeds, for, though the current through it will increase, the difference of potential at the ends of the filament will decrease, so that there is less power being expended in it. Some of the regulating resistance must consequently be cut out to keep the filament bright. The total E.M.F. used should be just sufficient to light the filament up to the required brightness for the deposition, without any extra resistance in the circuit. As the deposition proceeds and the resistance of the filament falls, resistance must be added to the circuit in order to keep the tempera- ture of the filament constant. More and more resistance has thus to be added until the added resistance is equal to one -fourth of the resistance of the filament when first lighted up. When this point is reached the resistance of the fila- ment itself will also be one quarter of its original resistance. If the deposition is continued beyond this point, the added resistance is gradually cut out of the circuit again, as the fila- ment henceforth gets duller instead of brighter as its resistance continues to fall. It is here assumed that the total E.M.F. is kept constant, and that the same number of watts are ex- pended in the filament throughout the process in order to maintain it at the constant temperature. As a matter of fact more power is required as the process proceeds, owing to the increase in thickness of the filament, so that the greatest amount of extra resistance required will not be so much as one-fourth of the original resistance of the filament. It is not often that filaments are flashed to the extent of reducing their resistance to one -fourth of the initial value. The filament must, therefore, be kept at the required bright- ness by the variable regulating resistance in series with it. When the ohm-meter or galvanometer indicates that the re- quired resistance is reached, the current is immediately turned off, and the filament is taken out and a fresh one put in. The process of flashing in coal gas at atmospheric pressure is not a good one. It is too rapid. The deposit is very rough, 54 THE INCANDESCENT LAMP AND ITS MANUFACTURE. and is apt to be sooty. A quantity of smoke is produced, and the glass jar is soon obscured. It is better to reduce the pressure by means of an exhaust pump. Instead of the exhaust pipe D (Fig. 16) being led into the open air, it is connected through a stop-cock to a mechanical exhaust pump. Between the stop-cock and the receiver is connected a syphon mercury vacuum gauge. When the filament has been inserted the vacuum cock is turned, and a vacuum is produced in the receiver, which should be ex- hausted until the vacuum gauge shows a difference in the height of the two sides of the mercury of about lin. The vacuum cock is then turned off and gas is admitted by the other cock until the vacuum is reduced to about Gin. dif- ference in level on the gauge. The filament may then be treated as before. The deposit will be less rapid and of a better kind, being harder and whiter and not so rough. If it is still too rapid, it can be made slower by reducing the pres- sure again after the gas has been let in. This process may also be varied by passing the gas over ether, gasoline, benzine, &c., but better results can be obtained by methods which do not use illuminating gas at all. The vapour of benzine, ether, gasoline, or other volatile hydrocarbons may be used alone. Whichever is used, the precise treatment will be different in each case, each hydro- carbon requiring to be used at a different pressure to produce the best results. A method in which the vapour of gasoline may be used may now be described. This liquid is much used in America for the manufacture of gas for illuminating and other pur- poses in situations where coal gas is not available. It is one of the lightest and most volatile liquids obtained from petroleum, having a specific gravity of only about O66. It has a very disagreeable smell, and is apt to vary some- what in its composition, some samples being more volatile than others. For this reason it is better to use " standard pentane," a liquid prepared specially for use in the Harcourt standard lamp. It is simply gasoline purified. It has no bad smell, and is always of the same quality. It has a specific gravity of about 0-63. It is, of course, a much more expensive liquid than most hydrocarbons, but it can be used almost up FLASHING. 55 to the last drop without waste, as it is entirely closed up and only admitted to the flashing chamber in exact quantities required for use. As it is extremely volatile it is necessary to keep it in a well stoppered bottle, with the stopper fastened down, or it will very soon disappear. The process about to be described, in which pentane is used, is capable of giving the filaments a very thin coating of deposited carbon. A circular brass base-plate, A (Fig. 17), turned perfectly true and free from flaws, is fixed to the bench. E \i E F D ^ v C _ , >> IB ~TT' FIG. 17. FIG. 18. Apparatus for Pentane Flashing. A small hole, Jth of an inch in diameter, passes through the centre. Into this hole is screwed and soldered the end of a brass tube B, which passes through the bench or table. Upon A is a flat rubber ring, C. The glass receiver D, which may be about llin. high and 2 Jin. in diameter, stands on the rubber ring. The whole of the clip arrangement on the top of the receiver can be lifted off. It consists of a vulcanite disc, E, underneath which is another rubber ring. Through the disc pass two brass rods, G G, ending in the spring ^clips H H, for holding the filament F. Four of these apparatus 56 THE INCANDESCENT LAMP AND ITS MANUFACTURE. may be fixed close to each other on the same bench, and can be operated by one person. Another arrangement is that shown in Fig. 18. Two metal rods, D D, pass through the base plate A, from which they are insulated, and terminate in the hooks E E. A glass shade, F, covers the whole and stands on the rubber ring on A. The filament is held in a clip such as that shown in Fig. 19. The metal parts C C are insulated from each other by mica, A, bound round with metal B. The clip is hung by the eye-pieces DD on the hooks EE (Fig. 18). Mica should be used for the insulation, as other insulating materials may not be able to stand the heating to which they will be subjected during a prolonged flashing. FIG. 19. Clip for Holding Filament during Flashing. Fig. 20 is a diagram of the vacuum connections. A A' B B' are the bases on which the receivers stand. The arrange- ment shown is such that A A' can be worked while B B' are being pumped. There is a syphon mercury vacuum gauge CCCC, about Gin. or Sin. high, for each receiver. This kind of gauge is necessary. A barometer tube gauge is too large and clumsy, and is apt to get broken when the vacuum is destroyed, owing to the impact of the mercury with the top of the tube. Moreover, the scale must be a movable one, so that it can be raised or lowered according as the barometer rises or falls, or the true state of the vacuum in the receivers cannot be known. A syphon gauge always shows the actual state of the vacuum, whatever the atmospheric pressure FLASHING. 57 happens to be. D D' are two-way cocks connecting A to A' and B to B'. E E' are three-way cocks which will connect A A' or B B' either to the air by F F, or by the three-way cock G to the pentane resevoir H, or to the vacuum pump K. The method of working is as follows : Filaments to be flashed having been put into the receivers on A A', the cock D being open ; the cock E, as shown in the diagram, connects A A' to G, and through G to the vacuum C-" Section of Three-way Cock. FIG. 20. Vacuum Connections for Pentane Flashing. pump, which may be at a distance in the engine room of the factory, and be driven by power. A vacuum is soon produced in the receivers on A A', and when it arrives at about Gin. pres- sure of mercury its progress can be watched on the gauge C C. It is essential in this method of flashing that all joints, pipes, and cocks be absolutely airtight. It might be supposed that arrangements such as shown in Figs. 17 and 18 would not be airtight. Standing with their own weight on the base-plates the glass cylinders would not be tight, but, as soon as the vacuum is turned on, the pressure of the air forces the cylinder 58 THE INCANDESCENT LAMP AND ITS MANUFACTURE. so tightly against the rubber rings that the arrangement does not leak, and, if properly made, it will hold a vacuum for hours. Of course, the arrangement shown in Fig. 18 with only one rubber joint is less likely to leak than the one (Fig. 17) where the filament is inserted through the top of the cylinder. The vacuum is applied until the difference in level of the mercury in the gauges is about half an inch. This is as far as most mechanical air-pumps will go in practice. The cock G is then turned, disconnecting the receivers A A' from the vacuum pump, and connecting them with the reservoir H, which has previously been partly filled with pentane. Pentane vapour now passes from H along the tubes into the receivers A A'. This is apparent by the increased pressure indicated by the gauges C C. The vapour should be allowed to enter until the gauges show Sin. difference in level. The cock G is then turned back again, connecting A A' with the vacuum pump, and the vacuum is again reduced to about IJin. by the gauges. The cock E is then turned so that A A' are discon- nected entirely. The cock E' is now turned so that BB' are connected through G to the vacuum pump. The condition of A and A' is now the same, and the cock D is turned, so that A is shut off from A'. Current is now applied to the filament in A. The increase in temperature of the gas in the receiver, produced by the heating of the filament, causes increased pressure, and the gauge shows a decreasing vacuum. The fila- ment is brought to a bright white heat, at which it is main- tained by means of the regulating resistance, until the indicator which is used shows that it has received a sufficient coating of carbon. The filament in A' is then treated in the same way. The cock D may then be opened, and E turned so as to admit air through F into A and A'. The filaments are then taken out and fresh ones put in. The same process is then repeated with the filaments in B and B'. While the flashing is going on in one side of the apparatus the other side is being pumped. This may appear a complicated arrangement of pipes and stopcocks, but with a little practice it can be worked very quickly. With the apparatus shown in Fig. 18, the electri- cal connections with the different filaments is made by FLASHING. 59 means of a switch, while with the other form (Fig. 17) one clip on the end of a flexible wire makes connection to the wires G G. By this latter method the chances of turning the current on to the wrong filament are less than with the switch arrangement, as the operator is apt to forget to put the switch over, whereas in the other arrangement the clip must be removed from the top of the apparatus before the filament can be taken out. A quicker method of flashing is to use a large receiver in which six or eight or more filaments can be flashed one after the other in the same vapour. This, however, is not a good plan, as the conditions are then different for each one. The only way to treat the filaments exactly alike is to do them separately. For this reason the cocks I) D' (Fig. 20) are put between the receivers, so as to prevent the gas and smoke from one entering the other as would otherwise happen. Immediately below A A' B B' is a bulb filled loosely with cotton wool to act as a filter, so that pieces of filament and dirt may not accumulate and block the pipes. There should also be another filter between G and the vacuum pump. The tubing and cocks may be either of metal or glass. The latter are more easily obtained and maintained vacuum-tight ; but they are, on the other hand, more liable to breakage. Con- nection between the cocks and tubes can be made by rubber "vacuum -tubing," the glass or metal tubes being first greased with vaseline. The brown coating which makes its appearance on the glass receivers during flashing can be easily cleaned off with a rag wetted with methylated spirit. The electrical connections will be in accordance with the method of indication adopted, either for flashing to resistance measured hot or cold, or any other system. Care should be taken in arranging the apparatus that the operator shall not be liable to receive shocks A mercury double-pole switch which is normally held off by a spring is recommended, so that the apparatus is only connected to the dynamo while the switch is held on. The current is left on continuously while the depositing takes place. The deposit of carbon produced by this method is hard and white, and of low specific resistance. By using a greater 60 THE INCANDESCENT LAMP AND ITS MANUFACTURE. pressure and quantity of vapour, pentane will give carbon varying from the hard white variety to a soft black one. The Author is not aware that any systematic experiments have been carried out for the purpose of discovering what is the best arnount of deposit to give the filaments. Any thick- ness can, of course, be given by prolonging the process. One use of the process of flashing, not yet mentioned, is that one size (diameter) of filament may be made to do duty for lamps which are to take currents varying considerably in amount from each other. Thus, a filament suitable for a 100- volt lamp may by flashing be made into a 50-volt of the same candle-power, the one size of filament being available for all lamps with currents falling within such range. It seems probable, however, that one particular ratio of thickness of deposit to thickness of filament will give the best results for all sizes of the same filament flashed by the same method. Experiments of this sort are, however, very tedious to carry out, as conclusions can only be arrived at by life tests of a number of lamps of each of the variations. The practical limit of the use of one size of filament for various currents is, however, determined by the time taken to flash. The longer the time the more costly will be the pro- cess, as fewer filaments can treated be by each apparatus and operator. The cost of a long flashing will soon mount up, and be greater than the extra trouble of making different sizes of filaments at first. Double the time of flashing means half the number of filaments, or twice the apparatus and number of operators. Provided that the receiver of the flashing apparatus is large enough to hold plenty of hydrocarbon vapour, the thickness of deposit is proportional to the time during which the fila- ment is lighted up. This is not correct for a very prolonged flashing, as the hydrocarbon vapour becomes impoverished and works slower and slower, but with a properly proportioned receiver it may be taken as being so within the time which can be allowed for the flashing. Time really fixes the limit to the amount of flashing. The pentane method described will produce a good and even deposit in from twenty seconds to one minute. The conditions can be arranged so that the deposit proceeds at the rate of about 0-OCOOlin. per second. FLASHING. 61 In thirty seconds there will be then a good deposit OOOOSin. thick. A coating of this thickness will be just as smooth as the filament upon which it is deposited. Prolonged flashing gives a rougher deposit which may have a varying emissivity, and, therefore, a thin deposit is better. It is most convenient in practice to give the same thickness of deposit for all sizes of filament. The same time will then always be taken in flash- ing, whatever the size of filament. The size of filament must, therefore, be varied according to the current it is to take. No one size should be used for filaments taking different strengths of current, except within very narrow limits. The E.M.F. required to flash filaments depends upon their length, thickness and specific resistance. For a 100-volt 16-c.p. filament from 200 to 300 volts may be required at starting, and as much as 1-6 ampere may be sent through the filament during the flashing when it will only take 0-64 ampere when finished. This great difference is due chiefly to the cooling effect of the convection currents within the receiver, which are, of course, absent in the finished lamp. When flashing at atmospheric pressure or under a liquid the difference is much greater. If the filaments are not properly carbonised before flashing, a much greater electrical pressure may be required to light them up in the first instance than that given above. As already mentioned, the method of flashing filaments to a certain resistance, even when it is accurately done to the hot resistance, does not produce uniform results in voltage unless the size and emissivity of the filaments are also uniform. There is, however, another method in which the size and emis- sivity do not matter within certain limits. The filaments are flashed directly to voltage, and the candle-power and resistance may take care of themselves. Unfortunately, however, the eye of the operator has to be depended upon for producing and main- taining the right temperature even more accurately than when flashing to resistance. The quantity of hydrocarbon vapour, and particularly the pressure, must be exactly the same for each filament. A voltmeter is connected so as to indicate the volts at which the filament is being run. The current is turned on, and the filament is brought as quickly as possible up to the required brightness. Suppose that the filaments are for 62 THE INCANDESCENT LAMP AND ITS MANUFACTURE. 100 volts and 16-c.p. : the voltmeter indicates, say, 200 volts. As the deposition proceeds the filament falls in resistance, but by means of the regulating resistance it is kept at the same brightness. The indication of the voltmeter is all the while gradually falling, and when it reaches, say, 170 volts the current is cut off. Now, provided that the conditions of temperature or brightness and the amount and pres- sure of the hydrocarbon vapour are the same in each case, it does not matter whether the filaments are 15, 16, or 17-c.p. ones ; if the process is stopped when the volt- meter indicates 170 volts they will be alike in voltage when made into lamps, 170 volts in the flashing receiver being the corresponding value for 100 volts in the properly exhausted lamp. The actual voltage at which the current is to be cut off must, of course, be found by trial in the first instance for the particular apparatus and conditions employed. The degree of exhaustion in the receiver must be accurately reproduced each time. The greater the pressure the greater will be the E.M.F. required. A similar method can be used for making lamps to work at the same current, an ampere- meter being used instead of a voltmeter. Hard, white, deposited carbon has only about one-tenth of the specific resistance of carbon made from amyloid, the resistance of the deposited carbon of that kind being about 350 microhms per cubic centimetre, and "that of amyloid carbon 3,500 microhms at a temperature of from 6 to 2 watts per candle power. At the ordinary temperature of the air the deposited carbon is about two and a-half times greater, and the amyloid variety one and a-half times only. It will thus be understood that the ratio of the cold resist- ance to the hot resistance of a flashed filament depends entirely on the ratio of the amounts of the two kinds of carbon of which it is composed, a much-flashed filament having a larger ratio than one only slightly flashed. The resistance of a filament measured cold gives no indication of its hot resistance unless that ratio is known. The most usual ratio is about two to one, though it is often above or below that amount. The specific resistance of arc-lamp carbons (cold) will vary from 13,000 microhms per cubic FLASHING. 63 centimetre to less than that of amyloid, according to the method of manufacture, moulded carbons having a higher specific resistance than squirted ones. Different kinds of carbon have different emissivities that is, they radiate heat and light at different rates, so that, at a given temperature, carbons of the same size but of different makes require different amounts of power expended in them in order to maintain that temperature. This curious property appears to be connected with the nature of the surface only, and has nothing to do with the material underneath the surface. It is well known that a hot polished copper rod, if covered with lamp black will cool much more rapidly than a similar rod not blackened. In the same way a hot lamp filament with one kind of surface will cool faster than a similar fila- ment with another kind of surface. A filament made of amy- loid carbon will cool faster than the same filament if it be flashed in pentane in the way described. Suppose that an amyloid carbon filament be made into a lamp and tested, and that it is found that at 4 watts per candle power it takes 84 watts and gives, consequently, a light of 21 candles. Let the same filament be taken out of the lamp and flashed a very little by the pentane process, and then be made into a lamp again and tested. Let it be run at the same temperature as before, and it will be found to give only 15 candles and take only 60 watts. Now that it is flashed and has a different kind of surface it loses heat much more slowly, and a less ex- penditure of power suffices to maintain it at the former tem- perature. It must be particularly noticed, however, that there is a corresponding decrease in candle power. Let it be run at its original candle-power (i.e., 21) and it will be found to take about 68 watts, equal, that is, to about 3J watts per candle, considerably less power than before flashing. Many people have supposed that results like this indicate that the flashed filament is a more efficient one than the unflashed. This, however, is not at all the case. The filament now giving 21 candles is at a higher temperature than it was when giving that candle power before being flashed. If it had not been flashed at all, it could have been equally well run at this tem- perature and efficiency, only it would then have given about 64 THE INCANDESCENT LAMP AND ITS MANUFACTURE. 28 candles and have taken about 90 watts. The filament has simply become less emissive : it radiates more slowly. The extent of surface of filaments will vary with different kinds of surfaces from 6,000 to 10,000 square mils per candle- power at the temperature of 4 watts per candle-power (equal to 100 to 170 c.p. per sq. in.). When run up to the tempera- ture at which the best carbon begins rapidly to disintegrate (0-8 watt per candle-power) as much as 1,900 c.p. per sq. in. may be reached. The crater of an arc-lamp carbon is said to give about 45,000 c.p. per sq. in. The change in emissivity produced by flashing gives an additional reason for carefully reproducing the conditions under which that process is carried out for each filament so that they may all come out alike. The actual amount of the error in the voltage of the lamps produced by the several causes mentioned will now be con- sidered. An error in the thickness of a filament is a serious one. Suppose that a filament for a 100-volt 15-c.p. lamp at 4 watts per candle-power (= 60 watts total) should be 6 mils in diameter, but that the conditions of flashing have not been quite right, so that the diameter comes out at 6*4 mils, an increase of 6-6 per cent, in diameter, and therefore in surface. The filament will now require 6*6 per cent, more energy, and will be increased a like amount in candle-power. It will take 64 watts and give 16 c.p. But it was flashed to the resistance for a 15-c.p. lamp (166 ohms), not for a 16-c.p. one. Conse- quently, it will take 103*2 volts and - 62 ampere, to give the right temperature. Or if run at 100 volts it will give only about 13*5 candles, and an efficiency of 4J watts per candle- power. It will be noticed that, though the increase in diameter is 6*6 per cent., the increase in voltage is only about half that amount, the error being divided between the E.M.F. and the current. The percentage error in voltage will be about one- half that of the diameter for ordinary variations. With regard to errors in emissivity due to the conditions of flashing being wrong, the effect is exactly the same as for errors in thickness, an increased emissivity having the same effect as an increase in diameter. A similar result will be FLASHING. 65 produced if filaments are mounted with wrong lengths. This is usually looked upon as a less serious matter than differences in thickness, as it is supposed to be easy to mount filaments 60 80 10Q 120 14O 160 180 200 Time of Flashing ; Seconds. FIG. 21. Curves showing Effect of Flashing on the Resistances of Filaments of Various Diameters. to the proper length. Suppose a filament is to be 5in. long. One per cent, is O05in. that is to say, it must be mounted with an error of not more than O025in. on each leg, in order 20 40 60 80 100 120 Time of Flashing; Seconds. 200 FIG. 22. Curves showing Effect of Reduction in Resistance, due to Flashing, on the Voltage of Filaments of Various Diameters. to be within 1 per cent. It is, therefore, very important to be exact in the matter of length. With errors of resistance the same kind of result occurs as with errors in surface. One per cent, error in resistance will F 66 THE INCANDESCENT LAMP AND ITS MANUFACTURE. produce only about one-half per cent, error in voltage. This must not be understood to apply to very large differences in resistance. For instance, a 50 per cent, reduction in resistance gives more than 25 per cent, reduction in volts. The actual 360 40 60 80 100 120 140 Time of Flashing, Seconds. FIG. 23. Curves showing Effect of Reduction in Resistance, due to Flashing, on the Current for Filaments of Various Diameters. effect on the volts and current of a gradually diminishing resistance is shown in the accompanying curves, Figs. 21, 22, 23. The resistance is supposed to be diminished by flashing, the thickness of the deposit of flashed carbon being FLASHING. 67 considered as proportional to the time of flashing, the rate of flashing being such that a 4-mil filament is reduced 50 per cent, in resistance in ten seconds. It must be understood that the effect of reduction in resis- tance alone is shown in these curves. The actual results of flashing, taking into consideration the change in resistance, emissivity and thickness, are fully dealt with in a subsequent chapter. It has been shown that the time of flashingjis an important consideration, and that it is better to flash always to about the same length of time, and to use filaments of different diameters which will permit of this being done, than to use only a few sizes of filaments and to vary the time of flashing. It is therefore necessary to consider the relation of the sizes of filaments to the candle-power, voltage and tempera- ture at which they are to be run. F 2 CHAPTER VI. SIZES OF FILAMENTS (UNFLASHED). A HOT BODY may lose heat in three ways : (1) by conduction to another body with which it is in contact ; (2) by convec- tion currents in the medium by which it is surrounded, or (3) by radiation. The filament of an incandescent lamp practically loses heat by radiation alone. Some is lost, no doubt, by conduction along the supporting wires, whence most of it is radiated at a lower temperature, a little being conducted away to the fastening outside the lamp. The proportion lost by conduc- tion in this way is insignificant, except, perhaps, in the case of large current small candle-power lamps, and need not, therefore, be considered. In a properly exhausted lamp the loss by convection will be practically nil. Radiation, there- fore, may be considered to account for the whole of the loss of heat. The rate of loss of heat of a conductor by radiation is pro- portional to its surface. When a constant temperature is maintained in the conductor, the rate of loss of heat must be equal to the rate at which heat is generated in the conductor. If heat is generated at a greater rate the temperature will rise, if at a less rate it will fall. If the same temperature is to be maintained in different conductors, the extent of surface of those conductors must be proportional to the rate of generation of heat within them. The rate of generation of heat in lamp filaments is measured in watts. Consequently, in order to maintain the same temperature in different sized filaments, the extent of surface of the filaments must be proportional to the watts expended 70 THE INCANDESCENT LAMP AND ITS MANUFACTURE. in them ; or, vice versa, the watts expended must be propor- tional to the extent of the surfaces of the filaments. The filaments must, offcourse, be in a good vacuum, and be of like emissivity. The relation of the,' power expended to the extent of radiat- ing surface is theVrulingTfactor upon which everything else depends in determining the size of the filaments. With fila- ments of the same emissivity this relation will be constant for any given temperature, no matter whether the filaments be long and thin or short and thick, or whether they are circular or flat, solid or hollow. As a rule, filaments are circular in section and solid, not tubular, and this kind will, therefore, be first considered. Circular Filaments. Let d = the diameter of the filament, I = length r = ,, resistance ,, ,, s = ,, surface ,, ,, c = ,, current ,, ,, and c 2 r = ,, power spent in the filament. Now, as the power spent'is proportional to the surface, we have c 2 r oc s ; the resistance r is proportional to the length I and inversely proportional to the square of the diameter, or I rcc V (unflashed filaments are, of course, being considered), and the surface s is proportional to the diameter, multiplied by the length, or s oc dl. Therefore, c 2 x - ^t * i I SIZES OF FILAMENTS (FLASHED). _Afv AFTER DIAMETER OF FILAMENT DCrOUe FLASHING ; MILS. \\J ^ \ \ 83 co I - a g *> CO 1 S t 14 .2 ^s : LJ c II * ; "S S Jl S3 .5 ** J3 p oSo|| 6 = ^< IM M r< % - - * = = = JL-8 o |""d Sw^ C o G2 84 THE INCANDESCENT LAMP AND ITS MANUFACTURE. Before commenting upon the curves, the method by which they were calculated will be explained. The filaments them- selves are of the same quality (i.e., specific resistance and emissivity) as those of the three lamps #, b, c, sizes for several shapes of which have been already given. This par- ticular quality, as already mentioned, is about the usual one obtained with filaments made from amyloid. For unflashed circular filaments it was shown (p. 71) that d = ac%, and that with this carbon the value of a is 10. In Figs. 24 and 25 the dotted curve A gives diameters for currents up to 3 amperes for this quality of unflashed carbon. We will now suppose that filaments of this same carbon are flashed, and that the flashing be done by the pentane process. The process is arranged as to the amount of pentane vapour in the flashing chambers, the degree of exhaustion, and the temperature to which the filament is heated, so that the deposit of carbon proceeds at the rate of 0-05 mil in thickness in 10 seconds ; that is to say, the diameter of the filament is increased by twice that amount, or 0*1 mil in 10 seconds. This is a convenient rate in practice, and one that can be easily obtained by most vacuum processes. From the dimensions found for the unflashed lamps a, b and c in the last chapter, the specific resistance of the carbon can be calculated, For convenience, we will take the re- sistance of one inch length of one square mil section. From lamp b we see that a filament 7*4 mils in diameter and 4*33in. long has a resistance of 156 ohms ; consequently, one square mil lin. long has a resistance of 1,560 ohms. The deposited carbon produced as described has a specific resistance of about one-tenth of that amount, or one square mil lin. long equals 156 ohms. The emissivity of this de- posited carbon is less than that of the carbon of the filament itself in the proportion of 14 to 10 ; that is to say, if an un- flashed filament at the temperature of 4 watts per candle- power gives a light of 14 candles, a filament of exactly the same extent of surface when flashed will give only 10 candles at the same temperature. With these data then, assuming any dimensions we like for the filament and any duration for the flashing process, we can calculate the current which it will take when flashed to main- SIZES OF FILAMENTS (FLASHED). 85 tain it at the required temperature (in the present case always that of 4 watts per candle-power), and we can then plot the results in curves. For example, take the case of a filament 10 mils in diameter. We want to find what current it will take, when flashed for, say, 80 seconds, to maintain it at the temperature of 4 watts per candle-power. Suppose it be Sin. long, what is its resist- ance ? We know that lin. length one square mil in section of this carbon = 1,560 ohms. Therefore, we get the resistance of this filament = 1>5 ? > * 5 =99-3 ohms. What candle-power will it be at the temperature of 4 watts per candle-power? We know (from the sizes on p. 71) that 6,300 square mils of surface of this carbon gives 1 c. p. at that temperature ; con- 10 XTTX 5,000 sequently. the candle-power of this filament = -- (TgOO = 25. Suppose now that it is flashed for 30 seconds. It will then have a coating of deposited carbon 0-15 mil thick. Its diameter will be increased by twice that amount and, therefore, will be 10-3 mils. The resistance of the deposited carbon is that of a tube 10 mils diameter inside and 10-3 mils outside and Sin. long. The sectional area = IT [ (r + t) 2 - r 2 ) ] = 4-79 square mils. The resistance of the deposited tube, then, is that of 4-79 square mils Sin. long, and is therefore 163 ohms (one square mil lin. long = 156 ohms). The flashed filament is therefore made up of the filament proper of 99*3 ohms, over which is a tube of flashed carbon of 163 ohms. The resultant resistance of the filament is conse- quently 61 '7 ohms. Now the candle-power of the filament before flashing would have been 25, but by flashing its emissivity is reduced in the ratio of 14 to 10. But it has also been thickened in the ratio of 103 to 100 Therefore its candle-power will now be and the power required will be 18*4 x 4 = 73*6 watts. XT o "' 73'6 -| i rvo Now c 2 = ~-= =1-193; therefore, c = 1-092 amperes. 86 THE INCANDESCENT LAMP AND ITS MANUFACTURE. Thus a 10-mil filament flashed for 30 seconds takes a current of 1/092 amperes, and the filament becomes 10'3 mils diameter. The same result is, of course, arrived at whatever the length of the filament is assumed to be, the length, as previously explained, having nothing to do with the current strength. By working out a number of cases in this way, curves such as those in Figs. 24 and 25 can be drawn. Having obtained the curves, they are useful in enabling us to do the reverse process at a glance. For instance, a par- ticular lamp is to be made, and its current must be so-and-so. The curves show at once what diameters of filament may be taken, and how much they must be flashed in order to produce the result. If it is desired to give the filaments a definite thickness of deposit, there is one diameter alone which will answer the purpose. Suppose that it is desired to give a coating OS mil thick. It takes 60 seconds to do this. Take the case of lamp b, a 100-volt 16-c.p. 0-64-ampere lamp. We find on the 60 seconds curve (Fig. 24) that we must take a filament 5-3 mils diameter to give this current. During the process of flashing it will become thickened to 5-9 mils, a fact which must be remembered in estimating its length. The length to O 1") be taken is given as before by the formula I = b -j-i & being the ct finished diameter, the value of b in the case of the flashed surface being 2-8 (i.e., the constant for the unflashed surface multiplied by 14/10, the ratio of the emissivities) ; 1 fi therefore I = 2-8 x = 7'6in. 5-9 7*6in. of 5-3 mils diameter filament must then be flashed for 60 seconds, and it will then be of the right resistance and surface. This, therefore, shows another way by which filaments may be flashed to the required resistance. No instruments are needed except a watch to show the time of flashing. Start with the proper size of filament shown by the curve, and flash it for the required length of time, and it is brought to the required resistance. Such a method, however, would not work well in practice owing to the difficulty of exactly repro- ducing to conditions each time. If, however, the right size of SIZES OF FILAMENTS (FLASHED). 87 filaments, as shown by the curve, are taken and flashed to their proper resistance hot, the time of flashing will be 60 seconds on the average. It will be seen by examining the curves that there is a wide range of possible sizes for the filament, according to the amount of flashing it receives. For the same lamp taking 0'64 ampere, a filament 3*7 mils in diameter flashed for two minutes gives the required result, or one of 7-65 mils flashed for only ten seconds will do equally well. It has, however, been pointed out in practice that 30 seconds is about as long a time as can be allowed for flashing. By taking the sizes given by the 30- seconds curve we get always, for any current, the diameter which will take 30 seconds to flash to its required resistance. On examining the curves, one thing that immediately attracts attention is that the curves for the flashed filaments D, E, F, G, H, J, cross the curve for the unflashed ones A, at some point. This is on account of the variation in the equivalent specific resistance of the flashed filaments. The topmost heavy curve B shows what the result would be if the filaments could be flashed so as to give the less emis- sivity of the deposited carbon, but without changing the resis- tance. In other words, it is the curve for a carbon of the specific resistance of the unflashed, but with the emissivity of the deposited variety. It is introduced in order to show the limit of diameter for any current beyond which it is impossible to go under any conditions with these two qualities of carbon. In the other direction the limit of the diameter is zero ; that is to say, the amyloid filament vanishes altogether, deposited carbon alone remaining. Thus the diameter selected for a 1-25-ampere filament at the temperature of 4 watts per candle-power may be anything from 12-9 mils down to nothing, according to the time of flash- ing. If unflashed carbon is used, the diameter, as shown by the dotted curve A, will be 11*6 mils ; if flashed for 10 seconds, it will be 12-3 mils ; for 20 seconds, 11*7 mils ; for 30 seconds, 11*15 mils ; for one minute, 9-7 mils ; for two minutes, 7-5 mils. The limit of zero diameter is reached (supposing such a thing possible) at the ten-minutes curve. It must be re- membered that the diameters shown by the curves on Fig. 24 58 THE INCANDESCENT LAMP AND ITS MANUFACTURE. are the diameters of the filaments before flashing, and that the actual finished diameters are greater according to the length of time they are flashed by one-tenth of a mil for every 10 seconds. In the limit when the diameter of the unflashed carbon becomes zero the actual diameter is + 6 mils, 6 mils being the increase in diameter produced in ten minutes. In other words, the filament is now entirely of deposited carbon, and 6 mils in diameter. Speaking, then, of the final or flashed diameter, the limits for 1-25 amperes are 12-9 mils and 6 mils, 6 mils being the diameter of a solid filament of deposited carbon alone. The lower heavy curve C gives the values for such material. It of course follows the d = a c$ law, the value of the constant a being 5-17.* In order to make the matter as clear as possible the second sheet of curves (Fig. 25) is given, showing the final or flashed diameters. It will be noticed that all the sizes lie between the limits of the heavy curves B and C. It is, however, the diameters of the filaments before flashing that the lamp- maker requires in the first instance. Looking at the curves (Fig. 25) we see for any current within what limits the final diameter of the filament for that current must lie. Thus, for two amperes the final diameter will lie between 8-2 and 17-7 mils. Theoretically, any dia- meter between these limits may be made to do for two amperes, according to the length of time of flashing, or no * The value of the constant may be found by taking any particular case and working it out, thus Suppose the filament of a solid deposited carbon to be Sin. long and 10 mils diameter, its sectional area = 78'54 square mils and its resistance, therefore, is -j * = 9'93 ohms. ' Its surf ace = 157,000 sq. mils. The surface per candle-power of the deposited carbon = 6,300 x 15 = 8,820 sq. mils; therefore its candle-power = zfpijrr = 17'8 f and the watts = 17'8 x 4 = 71 '2 8,820 therefore c2 = ! = .*. d 10 The sectional area of the filament = ?rr 2 = 78*54 square mils. We know that one square mil lin. long = 1,560 ohms ; therefore the resistance = * = 20 ohms. /3 = d*r = 2,000, SIZES OF FILAMENTS (FLASHED). 95 and therefore - - - 10, which is the value assigned to a at the first (p. 71). We are now able to find the value of the constant a for any carbon, if we know its relative specific resistance and emissivity to that of this carbon. Let us find its value by this method for the same carbon flashed so that its cold resistance shall be equal to twice its hot resistance. We know that by being flashed its emissivity will be reduced in the proportion of 14 to 10 ; consequently, a = 2x = l-43. As in this case there is always a fixed ratio between the sectional area of the deposit and that of the filament, we can find its equivalent specific resistance. It has been shown that the resistance of the deposit must be 0*6 times that of the filament. If, therefore, the resist- ance of the filament be 100 ohms, that of the deposit will be 60 ohms, and the combined resistance will be 37*5 ohms. But in being flashed the diameter of the filament is increased in the ratio of J7 (p. 91). Consequently, the resistance of an unflashed filament of the same diameter and length as the flashed one will be (1O.Q7\ 2 t^ \ = 85-6 ohms. Therefore the equivalent specific Io'o7/ conductivity has been increased by flashing in the ratio of 85-6 _ o.oo . 87*- consequently, P = 2,000 x -L = 876 ; therefore, a - J/f - ;J - 8-46, which is the value found on page 92. In the same way, by first finding the values for a and f3, the value of a can be found for any quality of carbon, or for flashed carbon which has a definite ratio between the sectional area of the filament and that of the deposit. 96 THE INCANDESCENT LAMP AND ITS MANUFACTURE. The following values for a, for filaments flashed so as to have a definite ratio between their cold and hot resistance, have been calculated in the same way. Conductivity of flashed Cold resistance. filament (hot). a ft a Conductivity of filament of equal diameter not Hot resistance. flashed (hot). 1-6 1-167 1-43 1715 10-62 1-7 1-362 1-43 1469 10-09 14 1-6 1-43 1250 9-52 1-9 1-9 1-43 1053 9-02 2-0 2-28 1-43 876 8-46 2-1 2-83 1-43 707 7-91 2-2 3-52 1-43 568 7-35 2-3 4-62 1-43 433 6-7^ 2-4 6-44 1-43 611 6-01 Curves for current and diameter for flashed filaments having these ratios of cold to hot resistance might be plotted on Fig. 25 along with the "time of flashing" curves D, E, F, &c., which they would cross. To save confusion one only (curve K') has been drawn. In many lamp factories, the sizes of filaments and the amount of flashing they receive are regulated entirely by guess and trial methods. Such methods, no doubt, answer the purpose in the end just as well as a calculation by a formula, or informa- tion obtained from a curve. But by a proper understanding of the effects produced by alterations in the diameter of the filament or the time or conditions of the flashing, the required results can be much more certainly and quickly obtained. Of course, it is impossible to get lamps all exactly right by any method. There are too many causes at work to upset the calculations, too many variable conditions to be simultaneously taken into account. Nine out of ten of the things to be considered may be right, the tenth one being wrong throws Ibhe combined result wrong. It is, however, only by taking everything into consideration that good results can be expected. . The curves and the constants given for use with the various formulas have about the values to be met with in ordinary practice. Carbons produced by different factories using very similar processes will, however, differ so much that it is not SIZES OF FILAMENTS (FLASHED). 97 possible to give any values which will be generally applicable. The exact values of the constants must, therefore, be found for each particular make of carbon and process of flashing. A good instance of the happy-go-lucky methods sometimes used in a factory was seen by the Author a few years ago. The filaments were here flashed in an apparatus, using an automatic cut-off, which stopped the current when it attained a certain value. This cut-out was a very crude affair, and was not at all particular as to when it went off. The flashing globes were allowed to get so black that the operator could not possibly gauge the temperature of the filament, and the consequence was that the cut-off would "act " just when the operator chose to turn the current up sufficiently. After this process the filaments were "selected" by measuring their resistance cold, a range of about twenty per cent, being allowed. The result was that three-fourths of the filaments were rejected. Of those coming within the limits of the resist- ance cold, perhaps one-third would be of the right resistance hot, which was always assumed to be half the cold resistance. The proportion of lamps of the proper voltage was, of course, extremely small, while the percentage of filaments which became lamps of the right voltage was microscopical. The factory above alluded to was, luckily, situated in a country where very badly-matched lamps could be sold as of the same voltage. It will be evident to the reader that the extreme care required in the production of uniform and good lamps necessarily adds to their cost. Lamps made without such care can be produced at a correspondingly lower rate. If the proper amount of care is taken during the manufacture, nearly all the lamps, when finished, will be close enough to the proper voltage and candle-power for them to be properly sold as such. When, however, such care is not maintained, the resulting lamps will, inevitably, be un-uniform, only a small percentage, possibly, being within the proper limits of voltage and candle-power. With the expiration of the lamp monopoly in this country, it is to be feared that the market will be deluged with inferior and badly matched lamps, which will be offered at a very low price. Consumers will, however, probably find that the higher priced lamps are the best and, in the long run, the cheapest. CHAPTER VIII. MEASURING THE FILAMENTS. THE filaments must, of course, be measured before mounting and flashing, and it is well to measure them again after flashing, to see that the increase in thickness is of the proper amount. The measurement may be done in several ways by means of micrometer gauges, or by throwing a magnified image of the filament upon a screen by means of a lamp and lens, or by the ordinary microscope method, which, however, 32nds. I .0312 3 .0937 '5 .1562 7 .2187 9 .2812 11 .3437 !3.4062 15.4687 1-8 .125 t-4 2. 5 18 .373 ICths. 1 .0625 3 .1875 5 .3125 7 .437 FIG. 26. Micrometer Gauge, for Measuring Diameter of Filaments. cannot be considered a factory method. Measuring with a micrometer can only be done with accuracy after a consider- able amount of practice. The best form of micrometer to use is that represented in Fig. 26, which can be read to the fifth part of a mil. It should be mounted on a support fixed to the table, so as to be about Sin. above the table, and, so that the screw head can be turned easily with the thumb and fore-finger of the right hand, with the fore-arm resting com- fortably on the table. The filament will be held very lightly 100 THE INCANDESCENT LAMP AND ITS MANUFACTURE. between the thumb and fore-finger, or between the first and second fingers of the left hand. Great delicacy of touch is required, as the filament may be squeezed to a considerable extent, and the measurement will then be wrong. Some micro- meters have a ratchet-head, so that they may be screwed up until the object to be measured is held just so tightly that the ratchet head slides round without closing the gauge any further. This arrangement is, however, not sufficiently deli- cate for filaments. They may be squeezed much too tightly before the ratchet comes into play. Of course a ratchet -head might easily be made to work with a very slight pressure, but the difficulty is to get one which will turn the micrometer screw without anything in the jaws (a considerable force being required to do this), and yet which will slip on the slightest extra resistance. A micrometer without the ratchet-head may be used, and it can be screwed up until it is felt that the fila- M IOOO" = width of the filament and t = thickness of the filament, the value of the constant will be 0-64 (l + -\ If w = 1-77 t, \ / then the flat measurement is also the mean. If many lamps with the same ratio of width to thickness have to be tested, a more convenient plan is to move the standard burner closer to or further from the photometer disc, so that, when measuring the lamp flatways, the mean candle- power is read directly on the photometer scale. If the ratio of width to thickness be not known, a measure- ment may be taken with the broad side of the filament towards the photometer, and another at right angles to it with the narrow side towards the photometer. If the candle-power in the former position = W and in the latter = T, then the mean candle-power will be 0*64 (W + T), provided that one side of the filament does not screen the other in making the test. The direction of the minimum light from a flat filament is at right angles to the thin edge. The direction of the maxi- mum light, however, is not that at right angles to the flat side, but at right angles to a line taken diagonally across the filament from opposite corners. In testing flat filament lamps it is possible, by turning the lamp, to find the position of maximum and minimum light. If the ratio of the width to the thickness be not known, the mean candle-power can, nevertheless, be found by a single measurement on the photometer. The lamp under test must be rotated until it is seen, by looking at the photometer disc, that it is in the position in which it gives the maximum read- ing. Let this maximum candle-power = H. The lamp is then turned through an angle j8, until it is seen that it is in the position of minimum candle-power. The mean candle-power is then equal to 0-64 H (sin /? + cos ft). The most accurate method of obtaining the mean horizontal candle-power is probably that of rapidly spinning the lamp, as the regularity of the shape of the filament does not matter. In various tests which have been published from time to time readings have been taken at every SOdeg. all round the 192 THE INCANDESCENT LAMP AND ITS MANUFACTURE. lamp. This gives a mean value sufficiently close to the real one for most purposes, though in the case of flat filaments the mean value obtained in this way may vary 3 per cent., according to the position of the filament that is to say, whether the flat side of the filament is set towards the photo- meter at angles of Odeg., SOdeg., GOdeg., &c., or 15deg., 45deg., 75deg., &c. In the former case the mean of the readings will be about 2 per cent, below the real mean, and in the latter case about 1 per cent, above. Such differences as these are, perhaps, hardly worth troubling about, except in the case of special tests. Before the lamps are tested they should be thoroughly cleaned. A dirty bulb will obscure the light very materially. After testing they must be sorted out according to their voltage. If they are loop lamps, they are finished except as to final cleaning and wrapping in paper. The cleaning is best done by wiping them with soft paper, first wet and then dry. This is much more effective than using a cloth. When finally cleaned, the lamps should not again be touched with the hand, or finger-marks will be left upon the glass. CHAPTER XIV. CAPPING. IF the lamps are to be capped the capping should be done after they have been tested. Lamps are fitted with caps for the purpose of providing a convenient and simple means of securely attaching them to a holder, and making the necessary electrical connections. The most usual form of cap in this country consists of a short length of brass tube which fits over the neck of the lamp and encloses the platinum wires. The platinum wires are usually soldered to copper wires, which in turn are soldered to the metal contact plates of the cap. The cap is filled up with some kind of plaster, which holds it to the lamp and secures the contact plates. There are pins on each side of the brass tube for securing the cap in the bayonet slot of the holder. There are numbers of different forms of caps designed to fit various patterns of holders. While it is not intended to discuss the merits of the different varieties, there are still a few points to be mentioned in connection with caps generally. Plaster-of-paris is generally used to fill up the space in the caps. The end of the lamp, as already explained hi the chapter on " Sealing-In," is formed in such a way that the plaster will have a firm hold upon it. Different samples of plaster-of-paris are found to vary very much. Some will set much more quickly and become much harder than others. Good plaster-of-paris sets quickly and expands in setting. The expansion in setting of some samples will swell the brass tube of the cap to such an extent that it will no longer go into the holder if it was of a fairly close fit before the plaster was o 194 THE INCANDESCENT LAMP AND ITS MANUFACTURE. put in. The addition of slaked lime to the plaster will, how- ever, reduce or prevent the expansion, according to the propor- tion used, which must be regulated by the quality of the plaster. The Author has also found that the addition of a small percentage of dextrine to the plaster-of-paris makes it very much harder. Another plaster sometimes used in caps is made of litharge, which is made into a paste with glycerine. This plaster, which sets very hard, is very heavy and more expensive than the plaster-of-paris. Current should never be put on the lamps until the plaster is quite dry. Plaster-of-paris takes at least four days to dry spontaneously. If the lamp is lighted before the plaster is dry a perceptible amount of current will pass through the plaster in the cap, and one of the copper wires will be electrolised. The Author has known cases where the fine copper wire has been eaten through and the connection with the lamp destroyed in this way. Plaster-of-paris can, however, be easily dried by heat, so that in a couple of hours' time after mixing it will conduct no current. Resin should be used as a flux for sol- dering the wires to the contact plates. When a lamp holder is discovered to be very hot, as is sometimes found to be the case, the heating is often caused by a current passing though the plaster of the cap, but is generally and often erroneously attributed to a bad contact between the holder and the cap. The trouble is sometimes due to the too abundant use of an acid flux in soldering, the excess remaining in the plaster. CHAPTER XV. EFFICIENCY AND DURATION. THE Author has adhered to the expression " watts per candle-power" in preference to that of "candles per watt," as, although the latter is the more correct, the former has the advantage of being the one generally adopted, doubtless be- cause it is more convenient. It is easier to quickly grasp the meaning of a certain number of watts, and perhaps a fraction added, than of a particular fraction of a candle-power. It will be readily understood from what has already been said that no lamp can be said to be of or to have a greater or less efficiency than another. One lamp may be run at a greater efficiency than another, and may then be said to be t a greater efficiency, but it possesses in itself no greater efii- ciency, as the second lamp can be equally well run at the higher efficiency. The higher state of efficiency simply means that the fila- ment is at a higher temperature. Any lamp may be run at any efficiency or temperature up to its breaking point. With incandescent carbon a certain temperature means a certain colour of light and a definite efficiency, either in watts per candle-power or in proportion of luminous radiation to total radiation. Whether the same thing can be said of all incan- descent substances does not yet appear to be definitely settled. Certain of the metallic oxides may, perhaps, prove exceptions. The well-known Welsbach gas burner seems to suggest such an exception, but the Author does not know of any tests upon this point. Certainly, to the unaided eye, the colour of the light of a Welsbach mantle is absolutely different from that of a gas flame, though the temperature of the particles of carbon o 2 196 THE INCANDESCENT LAMP AND ITS MANUFACTURE. in a gas flame might be supposed to be much the same as that of the Welsbach mantle. Certainly neither can be hotter than the gas flame. Possibly, however, the particles of carbon are very far below the temperature of the flame. These particles of carbon have a very great emissivity, and may, perhaps, radiate heat and light so fast that they never approach the actual temperature produced by the combustion of the gas. On the other hand, the Welsbach mantle, having a vastly inferior emissivity, is raised nearly to the actual temperature of the flame. The emissivity of the Welsbach mantle is extremely low. One which the Author measured gave when new less than eight candles per square inch of surface. The carbon in a gas flame gives many times this amount of light per square inch. Accurate tests with a spectro-photometer might settle this question. If the material of the Welsbach mantle is only at the same temperature as the carbon particles in a gas flame, then it must have the property of selective radiation, and may, therefore, be found to have a greater light- giving efficiency than carbon. In the case of using such oxides as the light-giving portion in electric lamps, one thing is apparent : that the extent of radiating surface will have to be very much greater than that of a carbon filament for the same amount of light if at the same temperature. The only way to ascertain the relative superiority of different incandescent lamps is to run them all at a constant pressure and all at the same efficiency at the start. They must then be tested at periods of, say, fifty hours at the same pressure as the original test, in order to find the diminution in candle- power. Such tests might be called " deterioration " tests. Thus, a lamp losing 10 per cent, in candle-power in 400 hours- is a better lamp than one losing that amount in 300 hours. Life tests, pure and simple, are worthless. A great many carefully-ascertained life tests of different lamps have at various times been reported. Such tests are, however, of no value unless the actual tests of candle-power of the lamp at different periods of its life are also given. It is also equally necessary to know within what limits the difference of potential was maintained throughout the run, as well as the length of time during which it was above or below the normal. EFFICIENCY AND DURATION. 197 All lamps fall off in candle-power when supplied with a con- stant number of watts. Lamps usually fall off in candle- power from the commencement of their active life, when sup- plied with current at a constant pressure. The deterioration in this case is, however, sometimes masked during the first hun- dred hours or so by the resistance of the filament falling. A greater amount of power is therefore supplied to the lamp, which may, in consequence, give even more light after a hun- dred hours than it did at the first. Such behaviour is, how- ever, the exception rather than the rule, and seldom occurs in flashed lamps. What amount of deterioration should be allowed before a lamp is cast aside it is difficult to say, and depends upon the circumstances of each individual case. As long as a lamp gives sufficient light in its particular situation there is no need to change it. A new lamp may be fixed over a writing table, and after a hundred hours may be found to be no longer bright enough ; it is therefore taken down, but it will do perfectly well in some other situation say, in a bedroom or passage for a further and longer period. The price of the lamp determines as much as anything the period at which it is thrown away. The cost of a candle-power hour for the actual amount of light given increases as the light of the lamp diminishes. In order that the total cost of maintaining a given amount of light may be a minimum, it is necessary to renew the lamps at certain definite times. The length of tune each lamp is used depends in such case upon the price of the lamp, the price of the power, and the rate of deterioration of the lamp. As the last of these conditions is variable even with lamps of the same make, it is impossible to fix the period under any conditions of price of lamps and power. Lamps in ordinary use cannot be tested every fifty hours, and it, therefore, happens that they remain in use as long as they give light enough. The length of time taken by lamps of the same make to deteriorate in candle-power by a certain percentage depends very much on their treatment. If the pressure is maintained constant, say within 2 per cent, on either side of the normal, the useful life will be longer than if the pressure is allowed to vary by 5 per cent. The pressure at which electricity is 198 THE INCANDESCENT LAMP AND ITS MANUFACTURE. supplied by central stations is anything but constant. The very best supplies are not to be relied upon to keep within 5 per cent, on either side of the normal, while some extensive supply systems do not keep within 10 per cent. The alternating systems seem to be more difficult to regulate than the direct current. Deterioration tests of lamps of the same make on the 180 60 100 c c 80 60 40 _ 108 112 92 96 100 104 Percentage of Volts. FIG. 64. Curve showing Percentage of Normal Candle-Power at any given Percentage of the Normal Volts. different supply systems would be very interesting and instruc- tive, and would go far towards settling the dispute as to the relative merits of the various systems in use. In order to show the importance of the constancy of the pressure, Fig. 64 is given. It shows approximately the per- centage variation in candle-power for any given percentage EFFICIENCY AND DURATION. 199 variation in the volts over a range of 10 per cent, above and below the normal. It will be found to be about correct what- ever be the temperature or efficiency of the lamp at normal voltage. The curve A gives the results when the filament is constant in resistance over the range of 10 per cent, in volts. The dotted curve B gives the result if the filament uniformly falls off in resistance by 5 per cent, for an increase of 10 per cent, in the volts that is to say, the resistance falls 1 per cent, for a 2 per cent, rise in volts, or 2 per cent, for a 4 per cent, rise, &c. As a rule the values lie nearer B than A. It will be seen that a slight increase in the pressure causes the candle-power to rise greatly, while a slight decrease means a great falling off in the light. The falling off in the candle-power of lamps is due to par- ticles of carbon being thrown off the filaments. This action reduces the light of the lamp in three ways. Firstly, the particles of carbon form a coating on the glass of the lamp, and, therefore, obscure the light given by the filament. Secondly, the nature of the surface of the filament is altered, and its emissivity is increased so that it is at a lower tempera- ture. Thirdly, the resistance of the filament is increased, so that it takes less current, and is again, on this account, at a still lower temperature. Thus, not only does the filament give less light, but a great proportion of what it does give is prevented from getting outside the bulb. In order to find out what proportion of the total loss of candle-power might be due to these different causes the Author made the follow- ing somewhat extreme test : A lamp was tested at about 2 watts per candle-power. It took 51-7 volts and gave 55 candle-power for a consumption of 107-7 watts. It was then run at that voltage until it was much blackened, and it was then tested again. At 51-7 volts it gave only 12 candles, and, owing to rise in resistance, it took only 100 watts. The total loss of candle-power was, therefore, 55 - 12 = 43, or 78 per cent. In order to find out how this loss was made up, the filament was taken out of the blackened bulb and put into a new one. The air was then exhausted to the same degree as before, showing a very good vacuum by the spark test. The lamp was then again tested. At 51-7 volts it took 100 watts, and gave 30 candles. It was 200 THE INCANDESCENT LAMP AND ITS MANUFACTURE. then run a little brighter, so as to take the original power of 107-7 watts, and it gave 42 candles and took 52-8 volts. The loss of light due to the blackened bulb was, therefore, 80 - 12 = 18 candles = 60 per cent. The loss due to increase in emissivity is given by the de- crease in candle-power, when the same amount of power was supplied to the filament. The loss due to change in emis- sivity is, therefore, 55 - 42 = 13 candles = 23-6 per cent. The loss due to change in resistance is given by the differ- ence in candle-power when the filament was supplied with the original number of watts, and when supplied with the original number of volts ; because, had the resistance not gone up, the original number of volts would have given also the original number of watts. The loss due to increased resist- ance is, therefore, 42 - 30 = 12 candles in 55 = 21-8 per cent. The rise in resistance is about 8 per cent. The loss due to change in emissivity, plus that due to change in resistance, therefore = 13 + 12 = 25 candles = 45-5 per cent. This result is also shown by the difference between the original candle-power and that at the original volts when the filament is in the new bulb, i.e., 55-30 = 25 candles. It is, therefore, apparent that the filament at the original volts gives 25 c.p. less than it gave at first : it gives only 30 candles, or 54'5 per cent, of its original candle-power. We have already seen that the blackened glass obstructs 60 per cent, of the light of the filament ; therefore, of this 54-5 per cent, of light given by the filament 60 per cent, is stopped by the blackened bulb. 60 per cent, of 54-5 per cent. = 32-6 per cent. The total re- duction in the light of the lamp is therefore made up in the following way : Less light given by filament, owing to Candles. Per cent. (1) Increase in emissivity 13 23-6 (2) , , resistance 12 21-8 Total (1) and (2) 25 45-4 (3) Stopped by blackened glass. 18 32-6 Total reduction 43 78-0 EFFICIENCY AND DURATION. 201 The greater part of the decrease in the light of the lamp is, therefore, owing to the filament giving less light. Of that which it does give only a part gets through the blackened bulb. Though the above test is perhaps an extreme one, the lamp having been very much overrun, the same kind of result undoubtedly occurs in lamps run at ordinary temperatures. The black deposit on the bulb of a lamp stops not only much of the light radiation, but also much of the heat. The result is that a blackened lamp often gets extremely hot. The effect of a fall in the resistance of a lamp is strikingly shown in the following test : An unflashed lamp had just been tested at 50 volts. It gave 12-5 candles, and took 0'82 ampere, and, therefore, 41 watts, and its resistance was 61 ohms. It was then accidentally run for an instant at 100 volts. The testing was being done on a 100-volt circuit, and the resistance in series with the lamp was momentarily short- circuited. The bulb was in consequence blackened, and the filament had lost its shiny appearance. It was, therefore, expected that a great diminution in candle-power would be observed at 50 volts. The resistance of the filament had, however, been so much reduced as to more than counteract the dulling effects. The lamp tested 50 volts, 1-03 ampere, 22 c.p., and, therefore, 51-5 watts and 48-5 ohms. The resist- ance had gone down 20 per cent., so that the lamp took so much more power than it did at the first that the increase in emissivity, which undoubtedly occurred, was not sufficient to prevent a great increase in the temperature of the filament ; and the blackening of the bulb, though great, was also not sufficient to mask the effect of the great increase in the light given by the filament. If a filament could be found whose resistance gradually fell as its emissivity increased and the bulb blackened, a lamp of uniform candle-power or of uniform watts per candle-power throughout its life might be possible. Such a lamp would not have so long a life as one of which the resistance increases, but it would have a more useful life, and would automatically end its career at the proper time. Unfortunately, the fila- ments which are usually found to fall in resistance do so during the first part of their life, when they should remain constant ; and when they should be going down they increase. CHAPTER XVI. RELATION BETWEEN LIGHT AND POWER. IT has already been stated that the light given out by an incandescent filament increases at a much greater rate than the power spent in the filament. The relation of the light to the power, or to the pressure or the current, has been the subject of much study and speculation. Many and various formulae have been given by different investigators. Some of them are nearly correct when applied to certain lamps, but are found to be quite useless when applied to others. No formula yet given appears to satisfy all cases. The reason probably lies in the imperfections of the lamps. The filaments of different lamps vary in their behaviour at different tem- peratures, and the vacuums also vary. In 1882 Prof. A. Jamieson showed that the candle-power of incandescent lamps was approximately proportional to the sixth power of the pressure, and, therefore, to the sixth power of the current, since the resistance of the filament when incan- descent does not alter much. The most recently published tests have been carried out, under the direction of Prof. Ayrton, at the Central Institution of the City and Guilds of London, and were published in part in Tlie Electrician of July 15, 1892. These tests are valuable, as they appear to have been carried out with great care to eliminate various errors, and the observations were made by three observers. It is shown by these tests that there is a definite relation between the logarithms of the candle-power and of the volts, amperes, or watts respectively of each lamp. As, however, the tests referred to are not carried above the temperature 204 THE INCANDESCENT LAMP AND ITS MANUFACTURE. of about 4 watts per candle-power, the Author introduces here some tests which he has made up to a much higher temperature." Lamp B. Lamp D. C.P. Volts. Amps Watts. Ohms aV C.P. Volts. Amps Watts. Ohms aV 5-3 29-0 1-13 32-6 25-7 5-15| 2-0 30-0 1-03 30-9 29-1 2-33 6-2 30-0 1-15 34-5 26-1 6-2 : 3-0 32-0 1-1 35-2 29-1 3-22 8'8 32-5 1-23 40-0 26-4 9-52 4-0 33-0 1-15 37-9 28-7 3-92 10-6 33-0 1-2$ 42-2 25-8 10-3 5-0 34-5 1-2 41-4 28-8 5-05 12-5 34-0 1-31 44-5 26-0 12-07 6-0 36-0 1-24 44-7 29-0 6-34 14-2 35-0 1-34 4tJ-9 26-1 1418 7-0 37-0 1-27 47'0 29-2 7'4 16'0 36-0 1-36 48-9 26-5 16-4 8'0 37-5 1-3 48-7 2b-9 7-97 177 36-5 1-39 50-7 26-3 17-65 9-0 38-0 1-33 50-5 28-6 8-5 22*2 38-0 1-45 55-1 26-2 22-0 10-0 39-0 1-36 53-0 28-7 10-0 26-5 39-5 1-49 58-8 26-5 26-8 12-5 40-2 1-4 56-3 28-7 11-68 35-4 42-0 1-58 66-4 26-6 37-4 15-0 41-8 1-45 60-5 28-6 14-5 44-0 43-5 1-66 72-1 26-2 45-0 20-0 43-7 1-51 C6-0 29-0 18-4 53-0 45-0 1-7 76-5 26-4 51-4 25-0 45-5 1-58 72-0 28-8 24-0 62-0 46-0 1-75 80-5 26-4 61-3 30-0 47-0 1-62 76 -5 29-0 27-6 71'0 46-8 1-78 83-5 26-4 66-8 35-0 48-5 1-H8 81-5 28-9 32-6 80-0 47-8 1-81 86-5 26-4 74-6 45-0 51-2 1-77 90 ''-> 28-9 44-2 106-0 50-5 1-9 96-0 26-6 100-0 f.0'0 52-2 1-85 1)4-5 28-2 48-9 124'0 52-0 1-94 10 1 -0 26-8 117-8 60-0 54-2 1-88 102-0 27-8 607 142-0 53-0 1-98 105-0 26-8 130-0 70-0 55-4 1-92 106-2 28-8 68-0 159-0 55-0 2-05 112-8 26-8 159-0 80-0 57-0 1-98 113-0 28 -8 80-0 177-0 56-0 2-07 116-0 27-0 174-5 PO-O 57-8 2-01 116-3 28-8 86-5 195-0 58-0 2'\ 3 123-6 27-2 210-0 100-0 59-6 2-08 124-0 28-8 102-0 212-0 60-0 2-2 132-0 27-3 2.i2-0 120-0 62-0 2-17 134-5 28-6 127-0 230-0 61-8 2-22 137-2 27 -H 296-0 140-0 64-8 2-2o 145-0 28-8 166-0 248-0 65-0 2-3 149-5 28-3 389-0 160-0 68-0 2-35 160-0 28-4 2U-0 "... 180-0 69-0 2-38 164-0 28-9 230-0 ... 200-0 72-o 2-49 181-0 2'l 300-0 250-0 80'0 2'7 21H-0 29-6 518-0 ... ... ... 295-0 90-0 2-8 252-0 32-2 icoo-o The lamps B, C, D, are of Edison- Swan manufacture, and lamp A is an unflashed amyloid filament lamp. The tests of A and C were, however, not carried very far. Fig. 65 shows the curves of watts and candle-power up to about 2 watts per candle-power, and Fig. 66 shows the results of lamps B and D run up to their breaking points. The dotted lines show the watts per candle-power. It will be noticed that both the curves (Fig. 66) turn over at the higher readings. This is caused by the disintegration of the filament and its consequent change in emissivity and re- sistance, and by the blackening of the bulb. The direction of the curves above the point where rapid disintegration begins depends, among other things, upon the length of time during FIG. 65. Curves showing Relation of Candle-Power and Watts for Four Lamps, A, B, C, D. 320 280 40 80 120 160 200 240 280 Watts. FIG. 66. Curves showing Relation of Candle-Power and Watts for Lamps B and D, up to their Breaking Points. 206 THE INCANDESCENT LAMP AND ITS MANUFACTURE. which the lamp is kept at each temperature while the obser- vations are being made. The curve of a lamp which is kept at 1 watt per candle-power for some time may, for instance, be made to bend over so much that it will never reach 0-9 watt per candle-power, as the lamp will break before arriving there. It will be seen that lamp B attains a much higher efficiency than lamp D. The observations at the higher readings in each case were, however, made as quickly as possible. Fig. 67 shows the relation between the logarithms of the candle-power and of the volts and watts respectively. Up to the point where the lamp begins to rapidly deteriorate it will be noticed that the observations in this figure lie nearly in straight lines sufficiently so to suggest that they ought to be exactly in straight lines. Assuming that to be the case, the expression C.P. = aV n , where C. P. = candle-power and V = volts, a and n being constants, would enable us to find the candle-power at any voltage when the values of a and n are known. In order to find the value of a and n for any lamp, two observations at least must be made. If c.p. = candle-power at the lower reading, ,, v = volts ,, ,, C.P. = candle-power ,, higher ,, V = volts then , -- . log and - v The observations from which the value of a and n are obtained must be made with the greatest care, and should be as wide apart as possible. A very slight error in the test will give a wrong value to the exponent n, and will make the calculated results quite wrong. In the case of lamp B, taking the readings of 30 volts 6-2 c.p., and 55 volts 159 c.p., the value of n is found to be 5-35 and a = 7-76xlO- 8 . In the same way for lamp D, at 39 volts 10 c.p. and 57 volts 80 c.p., 7i = 5-51 and a = 1-695 x 10~ s . RELATION BETWEEN LIGHT AND TOWER. 207 7 19 21 Log Volts, or Log. Watts. FEO. 67. Diagram showing Relation of the Logarithms of the Candle- Powers to the Logarithms of the Volts and of the Watts respectively of Lamps B and'D. 208 THE INCANDESCENT LAMP AND ITS MANUFACTURE. In Figs. 68 and 69 the dotted curves show the values cal- culated on the above basis. The crosses mark the observed values. Over the lower portion of the curves the observed and the calculated values lie very close together. In Fig. 69 it will be seen that the curve B, representing the actual obser- vations, definitely takes a different direction from B', represent- ing the calculated values at about 180 c. p., or at an efficiency / B ; * r ( / D/ / / ; i * V / / / / / i i i k / 1 1 / / 1 i i r / / / / ,' i / ' / 1 / i / / j / / L, / / / / V X 30 40 60 6( FIG. 68. Lower Part of Curves of Candle-Power and Volts for Lamps B and D. of 0*65 watt per candle. The curve D separates from D' at 100 candles, corresponding to an efficiency of 1*25 watts per candle-power only. This would seem to show that lamp D was an inferior one to lamp B. That great care is necessary in making the tests for a and n can be seen in the case of lamp B. If two observations only were made, and these were, say, those at 80 volts 6'2 c.p. and RELATION BETWEEN LIGHT AND POWER. 209 35 volts 14*2 c. p., the value of n would come out at 5-38, which is very nearly right, these two observations lying almost exactly on the straight line in Fig. 66. If, however, the tests were those of 32-5 volts 8-8 c.p., and 34 volts 12*5 c.p., the one reading of candle-power being a little low and the other a 320, 280 40 60 Volts 100 FIG. 69. Curves of Candle-Power and Volts for Lamps B and D up to their Breaking Points. little high, the value of n would be 7'9, and the calculated results would be quite wrong, these two observations lying one a little above and the other a little below the line in Fig. 67, where they are indicated by the dotted circles, while the observations from which the correct values are obtained are shown by the full circles. 210 THE INCANDESCENT LAMP AND ITS MANUFACTURE. It is probable that the relation of candle-power to watts is a more constant one among different lamps than that of candle-power to volts or amperes. In the latter case the degree of constancy of the resistance of the filament through- out the range of the test has an effect upon the form of the curve, whereas in the former case it has no such effect. Throughout this work the Author has adopted the efficiency of 4 watts per candle-power as the standard. This efficiency may be regarded as the highest ordinarily met with in lamps in actual use. At 4 watts a lamp looks bright as compared with a gas-flame. Although lamps are sold marked at 3-5 and 3, or even a less number of watts per candle, they very soon arrive at 4 watts when in use. The mean efficiency of any number of lamps in actual use in any installation, other than quite a new one, is probably quite as low as 5 watts per candle-power. At a temperature of 4 watts per candle-power, it has been estimated, by actual experiment, that the proportion of lumi- nous radiation to total radiation in an incandescent lamp is about 5 per cent, only, while the proportion in ths case of the arc lamp is about 10 per cent. In Fig. 66 it will be noticed that, by running lamp B up to its breaking point, an efficiency of 0*6 watt per candle-power is attained, equal to nearly seven times greater than at 4 watts per candle-power. It might, therefore, be expected that the proport on of luminous radiation to total radiation at this temperature was seven times that at 4 watts per candle-power, or 35 per cent., which would be far greater than that of the arc lamp. This, how- ever, is not at all the case. The candle-power is in no way a measure of the energy of the luminous radiation when comparing lights of different colours. Although by increasing its temperature we get an enormously increased efficiency as regards candle-power per watt expended in the filament, the ratio of the luminous energy to the total energy of radia- tion does not increase at anything like the same rate. The increasingly refrangible rays, which are produced as the temperature is raised, add little to the light as measured by the photometer, while they require comparatively a large amount of power for their production, Even under the RELATION BETWEEN LIGHT AND POWER. 211 >s. extreme conditions of O6 watt per candle-power the pro- portion of luminous radiation to total radiation of the in- candescent lamp probably falls short of that attained in the arc lamp. The ideal lamp is one from which the radiation is wholly luminous. The carbon incandescent lamp is, therefore, very far indeed from attaining the ideal standard. A more efficient lamp must evidently be looked for in another direction. The beautiful experiments shown by Mr. Nikola Tesla would seem to point to a direction in which it may possibly be found, but, as yet, illumination by means of vacuum tubes has not reached the efficiency of the incan- descent lamp, and is far behind it in point of practicability. In the preceding pages the Author has not touched upon the History of the Incandescent Lamp or the labours of the men to whom the world is indebted for bringing it to a state of perfection such that it has become an article of everyday use. In the first rank of these are the names of Swan, Edison, Weston, and Lane-Fox ; while following them come a host of others, who have materially assisted in making the lamp what it is to-day. In a book of this size it would be impossible to give credit to each for his share in the work, even if it could be ascertained. The manufacture of the incandescent lamp involves so many and various problems, and there have been so many workers in the field, that many inventions in connection with it have been independently made by different individuals. INDEX TO CONTENTS. PAGE Air Film in Lamps 162 Pumps 164 Amyloid, Composition of .. ... ... ... ... .. 14 Annealing Glass ... ... ... .- 123 Ayrton -Tests of Lamps ... 203 Bamboo, Filaments Prepared from ... ... ... ... 19 Blackening of Lamps ... .. 199 Blow-Pipe for Glass-Blowing 110, 111, 113 Air Supply for ... ... ... ... ... ... 115 for Pump-Room 169 Bulb Making from Pot Glass 106 ., from Tubing 116 , Moulds 107 Candle-power 190 of Flat Filaments 191 Falling off of 199 Increased by Fall of Resistance 197,201 Rise and Fall of, with Variation in Pressure . ... 198 Caps, Plaster or Cement for Attaching 194 Carbon, Disintegration of ... ... ... ... ... ...3, 199 Emissivity of Different Varieties of 4, 63 ., Resistance of 62 Volatilisation of 3 Carbonisation ... ... ... ... ... ... ... ... 5 Allowance for Shrinkage during ... ... ... 28 Crucibles for ... ... ... ... ... ... 26 Furnace for ...... ... ... ... ... 25 Necessity for High Temperature .. 30 ., Petroleum or Gas as Fuel for ... ... ... ... 29 Pyrometer for Use in 27,29 of Various Forms of Filaments ... ... 21, 23, 25 214 INDEX TO CONTENTS. PAGE Celluloid, Filaments Prepared from (Weston) ... ... 17 Cotton Thread, Filaments Prepared from (Swan) ... ... ... 7 Crooke's Process for Making Filaments ... ... ... ... 18 Cuprammonia Process for Making Filaments ... ... ... ... 18 Crucibles for Carbonisation ... ... ... ... ... ... 24 for Glass Making ~ ... ' 104, 105 Method of Packing Filaments in 24,25 Unpacking 30, 31 Curves Showing Effect of Flashing on Resistance of Filaments ... 65 Voltage ... 65 ,, Current ... 66 Diameter and Current of Filaments 82, 83 Variation of Candle-power with Variation of E.M.F. 198 ,, Connection Between Candle-power and Watts ... 205 Relation of Logs, of Candle-power to Logs. of. Volts or Watts 207 Candle-power and Volts Calculated and Observed 208, 209 Deterioration Tests ... 196 Diameters of Filaments see Filaments, Sizes of. Draw-plates for Cutting Parchmentised Thread ... ... ... 15 Drying Parchmentised Thread ... ... .. ... ... 13 Edison, Bamboo Filament ... ... ... ... ... ... 19 ,, Electro-plated Joint ... ... ... ... ... 35 Edison-Swan Pin-head Joint 38 Efficiency and Duration... ... ... ... .. ... .. 195 Emissivity of Different Varieties of Carbon .. ... ... 4, 63 Evershed's Wattmeter 188 Exhausting (see also Pumps) ... ... ... ... ... .. 131 Removing Gases from Filament ... ... ... ... 160 Air from Surface of Glass ... ... ... 162 Precautions against Moisture ... ... ... ... 167 Filaments, Errors in Thickness and Emissivity of ... ... ... 64 Length and Resistance of ... ... ... 65 Limits of Diameter for .. ... 87 Made to Serve for Different Currents by Flashing ... 60 Measuring ... ... ... ... ... ... 99 Produced from Bamboo, Edison ... ... ... 19 Produced from Cotton, Swan's Process ... ... ... 7 Cuprammonia Process (Crookes') 18 Westou's Process ... ... 17 Wynne and Powell's Process ... 16 Furfurol . 18 INDEX TO CONTENTS. 215 PAGE Filaments, Produced from Silk ... ... ... .. ... 19 ., Vegetable Parchment 19 Reduced to Required Resistance by Flashing ... ... 46 Sizes ,f, I'nflashed 69 Circular 70 Square ... ..73 Flat 74 , Hollow ... 76 Flashed ... 81 Spotty, made Even by Flashing ... ... ... ... 45 Flashing 45 Object of the Process... ... ... ... ... ...47 60 ,, Clip for Holding Filament during ... ... ... ... 56 E.M.F. required for ... ... ... ... ... ... 61 in Illuminating Gas at Atmospheric Pressure ... ... 51 ,. ,. Reduced 54 in Gasoline or Pentane Vapour ... ... ... ... 54 Vacuum Connections for ... 57 ., Regulating Resistance for Use in ... ... ... ... 53 Thickness of Deposit 60 Resistance Methods of and Difficulties ... ... ... 49 Time Method . . 86 Voltage Method 61 Furnace for Carbonising... ... ... ... ... ... ... 25 Oil and Gas 29 Glass Making 104 Geiss-ler Pump 134 Gimingham's Mechanical Mercurial Pump ... ... ... ... 176 Glass-Blowing 109 Annealing ... ... ... .. 123 ,. Blow Pipes for 110, 111, 113, 169 Bulb Making 116 Precautions with Lead Glass 112 Sealing-in 121 Making 103 Stoppers, &c., Ground .. ... ... .. ... ... 128 Tubing, Bulbs Blown from 116 ., ,. The Manufacture of 108 ., Method of Cutting 127 Induction Coil Test for Vacuum ... ... ... ... 173 Jamieson's Law of Candle-power Joint between Leading-in Wires and Filament, various types 34, 35, 36 ., Butt, Apparatus for Making ... 40 ,, of Edison -Swan Company... 38 216 INDEX TO CONTENTS. PAGE Joint, Carbon, Deposited from Gas ... ... ... ., ... 40 Liquid... ... ... ... 41 Deposited, Current required to Makp ... ... ... 42 ,, Resistance for Use in Making ... ... ... 43 Paste ;.. , 44 Socket, Apparatus for Depositing Carbon on ... ... ... 39 Kennedy's Pump ... ... ... . . ... ... .. 158 Lane- Fox Lamp, Early Form of t 36 Pump... ..' 136 Leading-in Wires, Compound ... ... ... ... ... ... 34 Platinum ... ... ... ... ... ... 33 Size of 38 Spiral Socket .. 37 Substitutes for Platinum 33 Tube 37 Length of Filaments see Filaments, Sizes of. Life Tests ... 196 Light and Power, Relation between 203 Maxim -Weston Lamp ... ... ... ... ... ... ... 36 McLeod Vacuum Gauge 170 Measuring ilaments ... ... ... ... ... ... ... 99 Mercury, Cleaning and Distilling ... ... ... . ... 174 Methods of Lifting, in Pumps ... ... ... ... 147 Precautions in Using ... ... ... ... ... ... 147 Micrometer Gauges for Measuring Filaments... ... ..,99, 100, 101 Mineral Substances in Conjunction with Carbon in Filaments . . 2 Moisture, Absorption of, by Filaments ... '... ... ... 32 in Mercurial Pumps ... ... ... .. .. ... 167 Mounting the Filaments on the Wires ... ... ... ... 33 Pentane Lamp ... .. ... ... ... ... ... ... 180 Pentane Vapour Used for Flashing ... ... ... ... ... 54 Photometer ; 181, 183 Plaster of Paris Used in Caps 194 Platinum as a Filament ... ... ... ... ... ... ... 1 Leading-in Wires ... ... ... ... 33 Substitutes for .. 33 Putnps, Mechanical ... ... ... ... ... ... 176 Gimingham's ... ... ... ... ... 176 Rotary 177 Mercurial 133 Geissler ... ... ... ... ... ... 134 . ., Joining Lamps to ., ... ... ... ... 168 INDEX TO CONTENTS. 217 PAGE Pumps, Mercurial, Kennedy 158 ,, Lane-Fox 136 Sprengel, Simple Form of.. 144 with Several Fall Tubes 145 with Air-passage Lift 148 Shortened Form of 150 > j ,, due to Steam ... 167 Swinburne .. ... ... ... ... ... 141 >, Factory Pattern ... ... ... 155 Toepler 138 > Weston ... ... ... ... ... ... 153 Pump-Room, Views of... ... 154, 156 Pyrometer, for Use during Carbonisation ... ... ... ...27, 29 Pyroxyline, Filaments prepared from (Weston) ... ... ... 17 Resistance, Effect of Fall of, on the Candle-power 197,201 Effect of Flashing on 65 Ratio of Hot to Cold 62, 90, 96 Specific, of Carbon 62 Variable, for Depositing on Joints ... ... ... 43 for Flashing 53 Sealing-in the Filaments ... ... ... ... ... ... 121 Shrinkage of Filaments during Carbonisation ... ... ... 22 Silk, Filaments prepared from ... ... ... .. ... ... 19 Spotted Lamps ... ... ... ... ... ... ... ... 125 Sprengel Pump, Simple Form of ... ... ... ... ... 144 with Several Fall Tubes 145 with Air Pressure Lift 148 Shortened Form ... ... ... ... .. 150 Steam's Pump 166,167 Surface Area of Filaments ... ... ... .. ... ... 64 Swan, Parchmentised Thread Process ... ... ... ... ... 7 ,, Strength of Acid used ... 11 . Drying the Thread ... ... 13 Cutting to Size ... ... 15 Lamp, Early Form of .. .. ... ... 35 Swinburne, Arrangements for Constantly Maintaining a Vacuum in a Pump 163 Method of Testing Lamps 182 Pump 141 ,, Factory Pattern I".". Temperature, Effect of, on Filaments during Carbonisation ... 30 Testing Lamps ... ... ... ... ... ... 179 Details of 183 Electrical Connections for ... ... ... ... 187 218 INDEX TO CONTENTS PAGE Testing Lamps, Light Standard for ... ... ... ... ... 180 Photometer for . 181 Wattmeter for... .. ^ 186 Vacuum... ... ... ... ...... ... ... ... 169 McLeod Gauge 170 by Induction Coil 173 Thompson, Prof. S. P., on Mercurial Air-Pumps .. ... ... 133 on Volatilisation of Carbon .. ... ... 2 Toepler's Pump 138 Torricellian Vacuum ... ... ... ... ... ... ... 133 Trotter's Micrometer Gauge 100 Vacuum in Lamps, Necessity for ... ... ... ... ... 131 Pumps see Pumps. Testing see Testing. Volatilisation of Carbon... ... ... ... ... ... ... 2, 3 Wattmeter, Evershed's 188 Necessity for 180 Welsbach Gas Light 195 Weston's Filament 17 Lamp 36 Pump 153 Wynne and Powell's Filament ... 16, 17 RETURN TO the circulation desk of any University of California Library or to the NORTHERN REGIONAL LIBRARY FACILITY SSg 400 Richmond Fieid Station i inix/prsitv of California Richmond, CA 94804-4698 "" prior to due date ^T^STAMPED BELOW LD 21-1 YC 69740 THE UNIVERSITY OF CALIFORNIA LIBRARY