January 17, 19 18] 



NATURE 



393 



Ix a paper which has appeared in the Proceedings 

 of the Tokyo Mathematico-Physical Society (1916, 

 p. 513, and 1917, p. 208) Prof. H. Nagaoka has taken 

 up the theory of the concave grating in order to deter- 

 mine the errors introduced when a wide grating is 

 used. He finds that the resolving power of concave 

 gratings is far less than that of flat gratings of the 

 same width. As, however, the utilisation of the whole 

 power of a plane grating necessitates the use of a 

 collimating telescope of large size, the best plan seems 

 to be to construct concave gratings of very small curva- 

 ture. The results of investigations of the structure of 

 spectral lines with the concave gratings hitherto avail- 

 able have been inferior to those obtained by interfer- 

 ence methods giving the same resolving power. But 

 the latter methods are in turn subject to the drawback 

 that the order of the spectrum for a particular con- 

 stituent observed may not be identical with that of 

 the principal line of the group under investigation. 



At the meeting of the Scottish Meteorological Society 

 on December 20 last, a paper was read on ground-ice 

 by Dr. John Aitken. It was pointed out that as ice 

 cannot begin to form unless the water is cooled 2° or 

 3° below freezing point, any radiation effect from the 

 bottom of rivers could never start the freezing there. 

 An experiment was described in which ice was formed 

 in running water by the action of radiation and cold 

 air. The ice so produced was in small crystals, or 

 frazil-ice, and was found attached to the bottom and 

 to obstructions in the stream, the same as observed 

 in rivers. This ice, when massed, was of a soft, 

 spon^v nature, like ground-ice. Observations made 

 where there is ground-ice show that the very slight 

 heating of the sun's rays soon causes it to loose its 

 attachment to the bottom and rise to the surface. It 

 is shown that this is due to the ice crystals slipping 

 when the temperature is Just above freezing point, and 

 adhering when it is just under it. The difference 

 measured on the thermometric scale is infinitesimal, 

 though the physical results are enormous. While 

 frazil-ice gives great trouble by adhering- to the hecks 

 of inlets at power stations, none has been experienced 

 from it adhering- to the guide-blades in the turbines. 

 This difference, it is pointed out, is due to the water 

 at the inlets being on the cold side of the freezing 

 point, while in the turbines it is just above it, owing 

 to its being under greater pressure and the ice melting. 



Prof. M. Tiffeneau informs us that the first volume 

 of the correspondence of Charles Gerhardt, the cen- 

 tenary of whose birth was celebrated by the Chemical 

 Society of France in December last, will be published 

 in a few weeks' time. This volume will contain fifty- 

 eight letters from Auguste Laurent and twenty from 

 Gerhardt, between the dates 1844 and 1852. It is 

 hoped that two other volumes of Gerhardt 's correspond- 

 ence will appear during this year. The complete work 

 will contain five hundred letters exchanged, for the 

 most part, with the chief chemists in Europe during a 

 period — 1837 to 1856 — which, from the point of view 

 of chemical science, was of the highest historical signi- 

 ficance. The publisher of the three volumes will be 

 M. P. Masson, 120 boul. Saint-Germain, Paris, and 

 the price will be about fifteen francs per volume, or 

 thirty-two francs for the set if this sum is subscribed 

 before the end of the present year. 



One of the difficulties with which railway mainten- 

 ance engineers have to contend is creeping of the 

 rails in a longitudinal direction, which necessitates 

 periodic rectification of the position of the rails after 

 the creep has taken place. Two papers were read on 

 • his subject at the Institution of Civil Engineers on 



NO. 2516, VOL. 100] 



January 8. In one of these papers Mr. H. P. Miles 

 describes investigations of this phenomenon made by 

 him for a period of five years in this country on a line 

 consisting of 850 track miles of main and branch lines, 

 over which various kinds of traffic passed. In the 

 other paper Mr. F. Reeves describes some simple 

 experiments he has carried out on pine, iron, and 

 rubber laths by causing loaded wheels to roll along 

 them. He concludes that creep is due primarily to 

 deformation of. the rail as the wheel passes over it, and 

 that the more violent the deformation, the greater will 

 be the creep ; thus creep is increased by increasing the 

 wheel load, and also by diminishing the rigidity of the 

 rail, either by reducing its section or by using a weaker 

 material. The weight of the wheel appears to be the 

 most powerful factor affecting the amount of creep. 

 Creep is accentuated by braking, and is greater down- 

 hill than up-hill, but is by no means absent on the 

 latter. Creep is always with the traffic. Creep can* 

 be resisted more or less completely by putting in 

 enough anchorage or resistance, and Mr. Reeves de- 

 scribes several such devices, including one of his own 

 design which is in use on the Buenos Ayres and Pacific 

 Railway. Many railway structures are affected by 

 creep, and their design should take it into considera- 

 tion. This requires special emphasis in the drawing- 

 office. 



OUR ASTRONOMICAL COLUMN. 



The Masses of the Stars. — The masses of all the 

 double stars for which sufficient data are available 

 have recently been calculated by Prof. H. N. Russell 

 (Popular Astronomy, vol. xxv., p. 666). The results 

 for the mean mass of a pair of stars, grouped accord- 

 ing to the spectral classes of the bright components, 

 are summarised in the following table, the unit being 

 the mass of the sun : — 



Physical From paral- 

 pairs lactic motion^i 



Spectroscoric Vimial 

 Spectrum binaries binarie 



No. Mass No. Mass No. Mass No Mass 



Bo-Bs 13 I7'5 8 104 36 7'i 



B8-A5 18 4-0 6 5-9 12 3-0 114 8*4 



F-G "giant" ... 34 3"9 37 8-i 



K-M „ ... 38 9-8 



F-F5 "dwarf"... 17 3"5 9 3'4 60 2-5 



F8-K0 „ ... 21 1-8 10 1-4 51 07 



K5-M ,, ... 4 0-7 8 10 



The first three groups present quite independent 

 data, but the fourth, though for the most part inde- 

 pendent, includes stars of the second and third groups. 

 The giant stars of all spectral classes are thus shown 

 to be nearly equal in mass, as they are in brightness. 

 Among the dwarf stars, however, where the luminosity 

 falls off rapidly with increasing redness, the mean mass 

 also falls off, but much more slowly. The masses of 

 the stars thus seem to be more closely related -to abso- 

 lute magnitudes than to spectral types ; that is, the 

 brighter stars are the more massive. This result is in 

 accordance with Prof. Russell's view that only the 

 more massive stars can attain great luminosity in the 

 course of their evolution. 



The Spectrum of o Camm Venaticorum.— It was 

 discovered by Belopolsky a few years ago that certain 

 lines in the spectrum of a Canum Venaticorum were 

 alternately visible and invisible, and the same observer 

 found later that such lines could be arranged in two 

 groups. In a brief report in Popular Astronomy 

 (vol. xxv., p. 656) it is stated that the spectrum has 

 been further investigated at the Detroit Observatory 

 by Mr. C. C. Kiess, who has obtained sixty-seven 

 photographs, and has determined the wave-lengths of 

 more than two hundred faint lines. The star is classed 



