College and Research Libraries and historical documents," is now one of the libraries of the University of Texas. This book is the autobiography of the man who built that collection. To say again: It is a curious book. One learns more of the basic facts of Mr. Par- sons' life from the jacket copy than from the book itself. It is a rambling, random account of travels, a downpour of name-dropping (of the author's real and bibliothecal ac- quaintances), and a gallimaufry of interest- ing and trivial anecdotes of the "But thereby hangs a tale" school of writing. It is over- larded with quotations, albeit some delight- ful ones. Too small a proportion of the book is devoted to a description of Mr. Parson's collection. For this we must look some day to a proper catalog from the University of Texas. And to say again: It is a charming book. Mr. Parsons' style is graciously Bourbon, as were his concept and his manner of collect- ing. His bookishness is deeply ingrained and his knowledge of books exceedingly wide. His memories of a by-gone New Orleans are delightfully rose colored. And some of his anecdotes are quietly tickling, particularly his telling of his gentle rebuff at the hands -or, rather, by the tongue-of that most underrated American humorist, President Calvin Coolidge. The Wonder and the Glory is a book for the bookish. The nearer one approaches Mr. Parsons' own bookishness, the more he will enjoy the book. He writes of himself: I had no "small vices." I did not smoke, drink, gamble or follow the races or the ladies. I had no time for Clubs or social groups. . . . Whatever time was left from law and the mak- ing of money was too little indeed to satisfy the real purpose of it all: the building of a collec- tion . . . , and the alluring study necessary to acquire an approximate knowledge of [books1 and of that endless sea of learning-bibliog- raphy. It was a great life, but it took a day of forty- eight hours to do it justice.-Richard Harwell, Bowdoin College. Vermont Imprints, 1778-1820. Compiled by Marcus A. McCorison. Worcester, Mass.: American Antiquarian Society, 1963. 597p. $15. (63-14992) This is the latest in a series of highly de- sirable lists of early imprints from specific JANUARY 1964 geographical areas; it is also a very welcome and worthy addition to the group of fine bibliographical studies that have come out of the American Antiquarian Society. This new volume identifies and when pos- sible locates 2273 books, pamphlets, and broadsides issued by the printers and pub- lishers who plied their trades in the Green Mountain State between 1778-date of the establishment of its first press in Dresden (now Hanover, N.H.) by the Spooners- and the year 1821. Bibliographical informa- tion includes author, his dates, title-page transcription, including place, publisher, and date, as well as pagination, size, locations of copies, and notes on copyrights. The work is obviously competently done. Compiler McCorison, who is librarian of the American Antiquarian Society, has com- pletely revised and included in the present volume the list of titles of Vermont Imprints before 1800 prepared by Elizabeth F. Cool- ey in 1937. By his own count he has elim- inated ninety-nine entries therefrom-sev- enty-four of which were newspapers and therefore covered definitively by Mr. Mc- Corison's someti.rpe colleague, the indefati- gable Clarence S. Brigham and therefore ex- cluded-and he has added 119 new items. Some rearranging of Cooley entries was also done because of revised authors or dates. Thus this list completely supersedes the Cooley list and is probably definitive. A sample check of McCorison's 1819 imprints, for example, against the recently-published volume of Shaw and Shoemaker's American Bibliography for the same year, shows some sixteen entries in the former which do not seem to appear in the latter. Some of the difference, however, might be accounted for by the different policies of inclusion used by the two works. McCorison includes as an appendix to his list a separate contribution of considerable consequence. This is a list of the more than three hundred individuals and firms that comprised the printing trades of Vermont during the period covered by the bibliog- raphy. Brief biographical sketches and ref- erences to further information are given for each. Herein are listed such important pub- lishers as Anthony Haswell and William Fessenden, such immigrant Scots and Irish bookmen as James Kirkaldie and John Hen- ry, and such printers as subsequently worked 67 the presses of the Western Country, includ- ing Samuel Vail and Matthew Lyon, also born in Ireland. Listed as well are book- binders, engravers, papermakers, and book- sellers. This is a valuable addition to the several directories of the early book trades of particular locales, most of which have been published by the New York Public library. The biographical sketches, inciden- tally, make interesting reading. Also given as appendices are a list of Vermont printers arranged by town, and a table correlating years and item numbers. As would be expected of a book of this kind, it is well indexed. As should be expected of a book of this kind, it is quite handsomely designed and manufactured by the Stinehour Press of Lunenburg, Vermont. We may hope that the publication of this important work will inspire similar studies of other geo- graphical areas.-D. K. Centralization and Documentation. Ar- thur D. Little, Inc. Cambridge, Mass.: The Author, 1963. 74p. This is the final report of a study sup- ported by the National Science Foundation. It considers the feasibility of centralizing fa- cilities for the storage and retrieval of scien- tific documents and makes the following recommendations to the National Science Foundation: 1. Do not support large-scale centralization of document searching facilities at this time. A large centralized facility drawing upon the current state of the art of document retrieval techniques could probably not achieve the main objective for which it was designed-provision of an effective, exhaustive, literature-searching capability to supplement efforts to prevent duplicated research or development invest- ments. Responsibility for showing that a pro- posed centralized facility would be feasible and would satisfy this objective must be borne by the proponents of centralization, employing quantitative evaluation techniques such as those we have developed. 2. Support the undertaking of a compre- hensive program to yield additional information and insight as to what the real informational needs of scientists and engineers are. Such a survey is a necessary prerequisite to the possi- ble support of centralized document searching facilities in the future, to insure that such fa- cilities will serve real functions, and that they will in fact be used. To be meaningful, the survey must be conducted with considerable imagination and insight. 3. Before undertaking extensive efforts to de- velop aids such as elaborate word thesauri for existing, partially centralized information re- trieval systems, investigate further the use of statistical techniques both for the automatic generation of thesaurus lists and for the auto- mation of some of the functions currently per- formed by human intermediaries. 4. To support such a program, test operate one of the medium-sized operating coordinate retrieval systems on a statistical associative basis. We feel that the state of the art of these associative techniques will permit such an un- dertaking, that a great deal could be learned from it, and that substantial benefits to the users of the system could quite possibly be realized. 5. For activities which are not concerned with exhaustive literature search operations, support centralization on an individual project basis, after cost effectiveness analyses have dem- onstrated-quantitatively-that adequate ser- vice levels and over-all benefits will accrue. It is heartening to see a study by an or- ganization that has been deeply involved in the past in installation of mechanized search- ing systems, as Arthur D. Little has been, that shows that they are willing to go where the facts lead. As they point out, the only "automatic" document searching system that has been applied on any substantial scale is the coordinate searching procedure and, "despite the use of high-speed digital com- puters, the searching logic employed in these most advanced systems is basically un- changed from that used in the earliest ap- plications . . . the systems are based on a purely mechanical attempt to match terms." The data analyzed indicate that in such sys- tems the indexing of all scientific literature by this approach would require using sub- stantially all meaningful scientific words as indexing terms, and even a file of half a million or so documents would require at least ten thousand index terms. A model is developed for study and evalu- ation of coordinate retrieval systems and applied to several collections. This shows that it will be difficult to obtain high preci- sion together with high recall ratios, and that even with an IBM 7090 (a very large-scale computer) the data processing cost for a large collection could become very great. These conclusions point up the need for (a) demonstration that cost and effectiveness 68 COLLEGE AND RESEARCH LIBRARIES