January 5, 191 1] 



NATURE 



301 



Chapter viii, is occupied with the method and pro- 

 '4ress of determining the motions of approach or re- 



ssion of celestial bodies b)' Doppler's principle of 

 changes of wave-length. 



The great field of solar observation is very efficiently 

 -immarised in chapters ix., x., and xi., including the 

 L\v results obtained by the use of the spectrohelio- 

 graph (a simple diagram such as is often used would 

 have been of value in rendering the explanation of this 

 instrument much easier); the phenomena of the pro- 

 iiiinences and reversing layer during a total solar 

 eclipse; the sun's rotation and that of the various 

 planets. In chapter xi. particular attention is given 

 to the spectra of terrestrial atmospheric phenomena 

 by the detection of special features in the solar spec- 

 trum at different altitudes. 



Chapter xii. is devoted to a short outline of the 

 meth(xls of investigating long wave radiations. 

 The inductive method of presenting the reason- 



g is very acceptable, and the beginner who has 

 lastcred the subject so far will be well equipped for 

 ntering on the more advanced branches of this 

 uricate section of sjjectroscopy. 



The concluding three chapters deal with the physical 

 • ctions of spectroscopy. The various systems of har- 

 monic laws found so closely to represent certain tv'pes 

 of spectra are well described. Perhaps in the pre- 

 sentation of the diagrams to illustrate these it would 

 have been preferable to adopt the same scheme of 

 orientation for the spectra. Thus in Fig. 46 the red 

 end is to the right, with all the lines of the series 

 converging to the left or violet; while in Fig. 47 the 

 red end is towards the left, and although the series 

 lines really converge to the violet as before, it is con- 

 fusing for a beginner to have to find that things are 

 all turned the opposite way. This is all the more 

 important from the fact that there are series actually 

 converging in opposite directions. The phenomena of 

 diffraction and general use of gratings for producing 

 spectra are next given, though necessarily condensed. 

 In a manual avowedly written to induce readers to 

 repeat the experiments, mention might well have been 

 made of the fact that excellent replicas of original 

 Rowland gratings, both plane and concave, are now 

 readily obtainable at a moderate price, as it is not 

 often that an original grating is available for general 

 experimental purposes. 



It will have been noticed that the description of the 

 application of the spectroscope has been almost con- 

 fined to its astronomical aspects; it should not be 

 forgotten that spectroscopic analysis is at present 

 playing an important part in the chemical and metal- 

 lurgical industries. 



Eight plates are given at the end of the book, show- 

 ing various representative spectra. There is also a 

 large coloured plate as frontispiece showing certain 

 elemental and other spectra. 



It will be evident from this summary that the book 

 under review should ser\e as a most useful intro- 

 duction to the study of spectrum analysis. It appro- 

 priately fills a position between the elementary primer 

 with little or no technical information and the more 

 formidable complete treatises which are admittedly 

 repellent to the beginner. 



NO. 2149, VOL. 85] 



THE THEORY OF METALLOGRAPHY. 

 Metallography. By Dr. Cecil H. Desch. Pp. x + 

 429. (London : Longmans, Green and Co., 1910.) 

 Price gs. 



"T^R. DESCH has evidently been at great pains to 

 J->' compile a work that shall give a fair idea of 

 the subject as a whole as it appeals to him, and he 

 is, above all, a theorist. It is a difficult work for the 

 writer to review, for two reasons : because it covers 

 practically the whole range of this enormous subject 

 and is therefore necessarily dogmatic on many matters 

 that, if disagreeing with the author, it would need 

 much space to discuss adequately ; and, secondlv, be- 

 cause he dismisses the whole Sheffield School thus : — 

 "This (the Osmond) hypothesis has been generally 

 accepted as the best expression of the known facts, in 

 spite of strong opposition from a (the Sheffield) school 

 of metallurgists . . ." although on pp. 363 and 

 364 we find rather contradictory- opinions, such as 

 "j8-iron was originally described by Osmond as a hard 

 variety of iron. It is more correct to say that it is 

 capable of forming solid solutions with carbon, which 

 become hard under certain conditions of cooling." 



The " eutectic-times " method for fixing the eutectic 

 composition was used by Arnold in his "Influence of 

 Carbon on Iron " (Proc., Inst. C.E., 1895-6, part i.), 

 although Tammann is credited with its first use in 

 1903 (p. 18). 



The author might with profit study "The Diffusion 

 of Elements in Iron," by Arnold and McWilliam, 

 I.S.I., 1899, No. I, instead of the preliminary an- 

 nouncement of 1898, which he quotes, and besides 

 further details on diffusion would find that these 

 authors used the quenching method then, in an 

 endeavour to judge of the condition of the carbon, 

 &c., at high temperatures. Also in connection with 

 the method of changing structure from that showing 

 Widmanstatten figures to granular, the author credits 

 the discover}' to Fraenkel and Tammann in 1908, 

 whereas the fundamentals of the matter were first 

 published by Arnold and Mc\Mlliam in Nature, 

 November 10, 1904, p. 32. 



A good account is given of the diagram of thermal 

 equilibrium, and on p. 32 it is pointed out that the 

 intermetallic compounds do not conform to our ideas 

 of valency. 



The sixth chapter is a good one on practical pyro- 

 metrj' and thermal analysis, but actual work on the 

 subject, and recent discussions have surely at last 

 made it plain, that the author is entirely mistaken in 

 his statement on p. 126 that " In accurate work on 

 the transformations of solids, however, one or other 

 of the difference methods is almost invariably 

 adopted." As recently as the Buxton meeting of the 

 Iron and Steel Institute, September, 1910, it was dis- 

 tinctly shown that in a o"2 per cent, carbon steel the 

 best workers by the difference method do not divide 

 the Ar, point, whilst those working with the present 

 modifications of Osmond's inverse-rate method divided 

 the A, point with ease, absolutely proving the supe- 

 rioritv' of the latter method. 



Chapters vii. to xiii. deal w^ith the preparation of 

 micro-sections; crystallisation of metals and alloys; 



