December r8, 1919] 



NATURE 



393 



radio-active changes, it would follow that the energy 

 given out in such changes must be greater in strong 

 gravitational fields than in weak. 



F. G. DONNAN. 



I'niversity College, London, December lo. 



i 



A Helium Series in the Extreme Ultra-Violet. 



_ In Nature for November 20 Prof. Lyman reports 

 his observation of a helium line at 1640-1, as well as 

 a weak one at 12151, close to the strong one 1216, 

 and refers them to orders 3, 4 in the series 

 4N{i/2'- l/w^j.. ]f the correct reading for the 

 strong line is nearer the 12151, the whole series 

 i» = 4 ... 8 is found in his list of ultra-violet lines 

 given in the Astro. Journ. (vol. xliii., p. 89, 1916). 

 The following is the list of observed lines with devia- 

 tions (obs. — calc.) from the calculated values, with 

 N = 109720 : — 



Order Intensity A d\ 



3 ... strong ... 1640-2 ... -034 



4 ... 10 ... 1215-1 ... -Q-ii 



5 ••• 5 ••• 1086-1 ... -I- 108 



6 ... 4 ... 10260 ... -I-0-66 



7 ••• 3 ... 992-0 ... -0-37 



8 ... 1 ... 9727 ... +0-59 

 The line fo84-9 is closer to the calculated with 

 dX= -0-11, but its intensity of 2 is not in step with 

 the others. In estimating possible errors, those of 

 standards as well as of observation have to be con- 

 sidered. With uncertainties also of formula, the 

 values of dX do not seem e.xcessive. 



W. M. Hicks. 

 Crowhurst, December 11. 



The Constitution of the Elements. 



It will doubtless interest readers of Nature to know 

 that other elements besides neon (see Nature for 

 November 27, p. 334) have now been analysed in the 

 positive-ray spectrograph with remarkable results. So 

 far o.xygen, methane, carbon monoxide, carbon 

 dioxide, neon, hydrochloric acid, and phosgene have 

 been admitted to the bulb, in which, in addition, there 

 are usually present other hydrocarbons (from wax, 

 etc.) and mercury. 



Of the elements involved hydrogen has yet to be 

 investigated ; carbon and oxygen appear, to u.se the 

 terms suggested by Paneth, perfectly "pure"; neon, 

 chlorine, and mercury are unquestionably "mixed." 

 Neon, as has been already pointed out, consists of 

 isotopic elements of atomic weights 20 and 22. The 

 mass-spectra obtained when chlorine is present cannot 

 be treated in detail here, but they appear to prove 

 conclusively that this element consists of at least two 

 isotopes of atomic weights 35 and 37. Their elemental 

 nature is confirmed by lines corresponding to double 

 charges at 17-50 and 18-50, and further supported by 

 lines corresponding to two compounds HCl at 36 

 .•ind 38, and in the case of phosgene to two compounds 

 C'OCl at 63 and 65. In each of these pairs the line 

 corresponding to the smaller mass has three or four 

 times the greater intensity. 



Mercurv, the parabola of which was used as a 

 standard of mass in the earlier experiments, now 

 proves to be a mixture of at least three or four isotopes 

 grouped in the region corresponding to 200. .Several, 

 if not all, of these are capable of carrying three, four, 

 five, or even more charges. Accurate values of their 

 atomic weights cannot yet be given. 



A fact of the greatest theoretical interest appears 

 to underlie these results, namely, that of more than 

 forty different values of atomic and molecular mass 



NO. 2616. VOL. 104] 



so far measured, all, without a single exception, fall 

 on whole numbers, carbon and oxygen being taken 

 as 12 and 16 exactly, and due allowance being made 

 for multiple charges. 



Should this integer relation prove general, it should 

 do much to elucidate the ultimate structure of matter. 

 On the other hand, it seems likely to make a satisfac- 

 tory distinction between the different atomic and 

 molecular particles which may give rise to the same 

 line on a mass-spectrum a matter of considerable 

 difficulty. F. W. .'\ston. 



Cavendish Laboratorx , Decemh)er 6. 



The Deflection of Light during a Solar Eclipse. 



The fall of temperature that may occur in the 

 higher strata of the atmosphere during an eclipse is 

 somewhat doubtful, but can scarcely exceed half a 

 degree. .An attempt was made to measure it directly 

 during the partial, but nearly total, eclipse in England 

 on April 17, 1912, but of the instruments sent up one 

 only was recovered, so that no comparison could be 

 made. 



On the average, the solar heat absorbed by the 

 earth's surface and the atmosphere during one dav 

 is capable of raising the whole atmosphere about 

 1-5° C, and, of course, about the same amount must 

 be lost per day by radiation. There is direct evidence 

 that the daily change of temperature as we know it 

 at the surface does not extend to more than i km. or 

 2 km., and from 2 km. to 20 km. the daily range can 

 scarcely reach 1° C. In these circumstances the fall 

 of temperature of the upper strata during an eclipse 

 must be small, say \° or i° at the outside. 



Our direct observations of atmospheric temperature 

 very seldom exceed 20 km., and above that height we 

 know neither the value nor the changes to which it is 

 subject. This is perhaps of little consequence since at 

 20 km. more than 94 per cent, of the whole mass has 

 been passed. 



It must also be remembered that the line of lowest 

 temperature will not be the axis of the shadow-cone, 

 but will lag considerablv behind it. 



W. H. Dines. 



Benson Observatory, Wallingford. 



The correction to the Einstein effect indicated by 

 Mr. W. H. Dines 's estimate of the depression of tem- 

 perature in the eclipse shadow is even less than 

 10"" radians or io-° seconds of arc. For a vertical 

 ray the deflection caused by a horizontal temperature- 

 gradient of dft° C. in dx cm. is approximately 

 , . ij (i log. d 

 ax 



where (^- i),. = 28 x lO"-' is the refractivity of air at 

 normal density, and H is the height of the homo- 

 geneous atmosphere. For two rays at a mean distance 

 in the atmosnhere of a kilometre apart — a liberal 

 estimate — the difference in deflection would be in c.g.s. 

 units 



28 X 10- ' X H X 10" . (i" log 6jdx-. 



If the shadow mav be considered to have a radius of 

 500 km., then d" {o^O/dx' would be of the order of 

 10- ". From this the initial statement follows. 



That is, assuming 9H/3-r = o. But more simply 

 conveniently, and accurately the bending of a vertical 

 ray can be Qxpressed in terms of the surface pressure- 

 gradient because the refractivity is simply proportional 

 to the density. So that five barometers, one at the 

 eclipse station and four distributed around it, would 

 yield the corrections for a single ray. The bending 



