344 



NATURE 



[January ii, 19 12 



and archteoiofry ii) the system of cross-refer- 

 ences which has been adopted throughout. One 

 obvious Rain has naturally been to avoid repe- 

 tition and inconsistency; but a further ad- 

 vantag*^ consists in the fact that the reader is enabled 

 to follow up various subsidiary lines of study in cog- 

 nate articles treating the subject in greater detail or 

 from a different aspect. We have already referred to 

 the article on y^gean civilisation ; the reader, if he 

 so desires, may find further information in other im- 

 portant articles on Crete, Mycena;, and the Troad ; 

 while the products of /ligean art are treated under 

 the more technical articles. By such a system of 

 cross-reference the value of the " Encyclopaedia," not 

 only as a work of reference, but also as a subject 

 for detailed study. Is vastly increased. 



There is one point on which we have not touched, 

 but which jxrhaps distinguishes more than any other 

 this edition of the ' Encyclopaedia " from its prede- 

 cessors—the number and beauty of the illustrations. 

 Many of the old line blocks and wood engravings have 

 naturally still great value, and where they have not 

 been rendered obsolete or out of date, they have been 

 retained; but they are supplemented by a wonderful 

 series of half-tone plates, arranged clearly on both 

 sides of the paper in order to save space, and con- 

 taining, in some cases, as many as twelve or sixteen 

 separate blocks to the page. For beauty of half-tone 

 illustration it would be hard to beat the series of 

 plates to the article on Greek art, while those to the 

 article on gems, with their admirably clear classifica- 

 tion, show what modern scientific illustration can 

 achieve. 



In the short account we have given of a single 

 aspect of this great work, we have not attempted to 

 critic.se any article in detail, though in the course of 

 our reading we have noted one or two slips. To 

 select them for special mention from a work of such 

 magnitude would be unfair, as it would tend to create 

 a wrong impression. That so high a standard of 

 accuracy should have been achieved is an eloquent 

 testimony to the devotion of the editor, and of the 

 distinguished band of specialists whom he numbers 

 among his contributors. ' L. W K 



THE DATA OF PHYSICAL CHEMISTRY. 

 Physico-chemical Tables, for the Use of Atialysts 

 Phystctsts, Chemical Manufacturers, and Scientin'c 

 Chemists. By John Castell-Evans. Vol ii 

 Physical and Analytical Chemistry. Pp xiv + 

 549-1235. (London: C. Griffin and Co. Ltd 

 191 1 •) Price 36s. net. 

 'X'HE second and concluding volume of Mr. Castell- 

 J- Evans's work on physico-chemical constants is a 

 veritable monument to the industry- and perseverance 

 of the author, who unhappily has not lived to reap 

 the reward of his labours. The user of the tables 

 may well be appalled at the magnitude of the task 

 undertaken by a single compiler to bring out two 

 volumes covering in detail the whole domain of 

 physical chemistry and running in all to more than 

 twelve hundred pages. 

 To review such a work adequately is almost impos- 

 NO. 2202, VOL. 88] 



sibte without keeping the book in hand for a 1< 

 time than is desirable. The reviewer's criticisms 

 therefore be taken as indicative of the directioi 

 which a future L*dition might be improved, r 

 than as the results of a detailed examination ; it si 

 also be noted that in the circumstances the revi 

 has refrained from raising debatable questions. 



The distinguishing feature of the book is utr 

 doubtedly the very large amount of labour put int( 

 the calculation of the numerous tables, a great nur-' - 

 of which are quite new and contain data only ol 

 able elsewhere from many scattered sources. I'M 

 example, the collection of data on viscosities and oi 

 densities of liquids and their variation with temper^j 

 ture are most useful and complete. To the analytici( 

 chemist and assayer the two hundred pages of tabid 

 for shortening analytical computations, for the corn. 

 parison of volume and weight f>ercentages, an^ 

 the conversion of the various systems of units emj 

 in such estimations, will be very valuable. 



In the compilation of this second volume ui t 

 even to a greater extent than in the first, the a 

 seems to have allow-ed his extraordinary desir^ 

 high numerical accuracy to lead him to giv- 

 values of constant in many cases to far more fii 

 than can possibly have any significance.' 



As a sample physical measurement of the hu 

 class under the test possible conditions may be 

 sidered the comparison of two similar plati; 

 iridium kilograms with all the refinements of a t: 

 position balance, a "complete" set of weighings 

 the utmost precautions. Under these condition- 

 thousandths of a milligram may be said to 

 some significance. Similarly in length measurer 

 the difference in length between two similar star 

 metres of the highest class may be determined 

 few hundredths of a micron, and the height 

 metre column of mercury to a few thousandths 

 millimetre. The last significant figure in the !:.. 

 cases is i/io*, i/io*, and i/io* of the whole respec 

 lively. These are the most favourable cases, but 

 seems unlikely that even with the best of methods ai 

 interval of temperature can be measured with greate 

 precision than about one part in a hundred thous.'ind 

 Besides, if instead of the mass or length chosen 

 the examples an odd multiple of a fundaments 

 standard be taken, the precision attainable may be onl 

 one-tenth of that given or even less. 



But Mr. Castell-Evans gives many instances c 

 eleven and some even of fourteen significant figure 

 among his constants and conversion factors, wherea 

 for the purpose of avoiding mere errors of oomputs 

 tion the retention of one, or at most two, additions 

 places beyond the usual limits would have been ampi 

 sufficient. 



While dealing with this question, it may be r« 

 marked that a physicist familiar with the modern pr< 

 cision measurements of physics, but whose menti 

 estimate of the kind of accuracy now attainable 

 chemical work was obtained only from Mr. Castel 



1 The writer is informed ihat this retention of an unusually large nuinb> 

 of figures was not due, as might have been supposed, to th« use of a calc 

 lating machine whereby the extra digits were obtained »-ithout any v€ 

 serious increase of labour, but that the numerical work was performed oyt 

 author without the aid of any machine. 



