NATURE 



[March 4, 1920 



their own way, with such advice or assistance as may 

 be asked for, the problems entrusted to them. 



The appointment of a head for each department of 

 science with the powers of a dictator would be the 

 surest means of encouraging mediocrity, and of warn- 

 ing off just that type of original thinker and indepen- 

 dent investigator whose services would be of inestim- 

 able value to the State. It may be contended that any 

 State scheme, whether concerned with routine duties 

 or original work, must be under some central direc- 

 tion, but there is no reason why the direction should 

 be of such a kind as would be tantamount to asking 

 every researcher to place himself, body and soul, under 

 a dictator. A. C. Seward. 



Botany School, Cambridge, February 26. 



The Constitution of the Elements. 



In continuation of my letter on the above subject in 

 Nature of December 18, 19 19, several more elements 

 have been subjected to" analysis, yielding interesting 

 " mass-spectra.'* 



Argon (atomic weight 3988 Ramsay, 39-91 Leduc) 

 gives a very strong line exactly at 40, with double 

 charge at 20 and triple charge at 135. The last line, 

 being closely bracketed by known reference lines at 

 13 and 14, provides very trustworthy values. At first 

 this was thought to be its only constituent, but further 

 photographs showed an associated faint line at 36. 

 This has not yet been proved an element by double 

 and triple charges, as the probable presence of OH 2 

 and the certain presence of C prevent this, but other 

 lines of reasoning make it extremely probable that this 

 is a true isotope, the presence of which to the extent 

 of 3 per cent, is enough to account for the fractional 

 atomic weight quoted. 



Helium was compared with 0++ (8) by a special 

 system of bracketing, and directly with C++ (6) by 

 extrapolation. Both methods give its mass as 4, with 

 an accuracy of 2 or 3 parts in 1000. 



By the same methods H3, H2, and H, all give con- 

 sistent results for the mass of the hydrogen atom as 

 I -008 within experimental error, agreeing with the 

 value given by chemical analysis, and, incidentally, 

 confirming- the nature of H, beyond doubt. These 

 three lines are the only ones diverging from the whole 

 number rule to a definite and measurable extent. 



Nitrogen is apparently a "pure" element, its doubly 

 charged atom being 7 exactly. 



Krvpton (atomic weight 8292) has no fewer than 

 six constituents : 78, 80, 82, 83, 84, and 86. The 

 last five are strong lines most beautifully a>nfirmed 

 by double- and triple-charged clusters, which can be 

 compared with great accuracy against A (40) and 

 CO (28). These reference lines obliterate one of each 

 group, but not the same one. The 78 line has not 

 yet been confirmed in this wav owing to its faintness, 

 but there is no reason to doubt its elemental nature. 

 Krvpton is the first element giving unmistakable 

 isotopes differing by one unit only. 



The partial pressure of xenon (atomic weight 130-2) 

 in the ^as used was only sufTficient to show its singly 

 charq-ed lines clearly. These appear to follow the 

 whole number rule, and rough provisional values for 

 the five made out may be taken as 128, 130, 131, 133, 

 and 13.1;. 



Further examination of the multiplv charged mer- 

 cury clusters indicate the probability of a strontf line 

 at 202, a weak component at 204. and a strong band 

 including 197 and 200, unresolvable un to the present. 



F. W. .\STON. 



Cavendish Laboratory. Cambridge, 

 Februarv 2f. 



NO. 2627, VOL. 105] 



Deflection of Light during a Solar Eclipse. 



Prof. Anderson has suggested in Nature that the 

 apparent displacement of stars observed during the 

 solar eclipse may be ascribed to an unusual form of 

 refraction in the terrestrial atmosphere. The discus- 

 sion which has followed shows some lack of agree- 

 ment as to the importance of such a refraction effect. 

 I wish to suggest that it might, perhaps, be possible 

 to form an estimate of the magnitude of this effect 

 by making measurements of the apparent diameter of 

 the moon during the eclipse. Star photographs would 

 seem to be somewhat unsuitable, although one dia- 

 meter of the moon may leave a clear enough trace on 

 the plates (a diameter at right angles to the apparent 

 motion of the moon relative to the stars). It should 

 be possible, however, to obtain sharp silhouette images 

 of the moon on plates devoted to this particular pur- 

 pose ; perhaps such photographs are already avail- 

 able. The nature of the clockwork drive needed is 

 dependent on the necessary exposure, and need not be 

 discussed. 



J. A. Orangk. 



Mr. Orange's point is, of course, that we should 

 use the one object in the field of which the light has 

 not been through the sun's gravitational field in 

 order to get rid of the Einstein disturbance ; also 

 of the suggested refraction by gases near the sun. 

 I have talked the matter over with Mr. C. Davidson, 

 who agrees with me that nothing is to be done with 

 existing photographs in this direction — the exposures 

 were too long, and the moon's limb too ill-defined; 

 but it is possible that in future eclipses short ex- 

 posures, given specially for the purpose, might vield 

 something of interest. The chief difficultv is that we 

 do not know the moon's dark photosjraphic diameter. 

 It cannot be assumed equal to the bright photogfraohic 

 diameter, for irradiation (and other similar actions) 

 go in the reverse direction. 



A. C. D. Crommeun. 



55 Ulundi Road. Blackheath, S.E.3, 

 Februarv 28. 



Perimeter of an Ellipse. 



The following approximate formula for finding the 

 perimeter of a fairly flat ellipse may be found prac- 

 tically useful. .Suppose a=i, then the length of a 

 quadrant of the ellipse is nearly 



l+o-6<^", 



where a is the major, and b the minor, axis. The 

 formula works best from about b = o-2 to ^=0-5, after 

 which the formula of Boussinesq is more accurate, 

 viz. 



^{t(l+^)-v/^}. 



But the formula I give is for practical purposes 

 quite satisfactory up to b=o-6, the relative error never 

 being large. It does not work if the ellipse is 

 nearly circular. Boussinesq's formula is of no use 

 if the ellipse is flat. 



Other more accurate formulae could be given, but 

 the above has the advantage that it can be calculated 

 very rapidly, and, within the range mentioned, I doubt 

 if higher accuracv is ever required in practice. 



R. A. P. R0GER.S. 



Trinity College, Dublin, 

 Februarv 16. 



