52 



KEPORTS ON THE STATE OF SCIENCE. — 1916. 



and is justified. The maximum in the other direction near A =20° 

 is only represented by the exceptional Taihoku record from which no 

 conclusion can be drawn. 



(y) Czernowitz, Vienna, Potsdam, and Gottingen all show a double 

 record near S, one member of which can be reasonably identified with 

 S and the other with the phenomenon Y mentioned in the last Eeport. 

 These four cases of double record ai'e specially valuable as a guide to 

 the others which only give one constituent, and it is easy to under- 

 stand why this should generally be the earlier one. But it must be 

 admitted at once that the explanation of Y given in the last Eeport 

 breaks down. It cannot be a ' polychord,' at any rate not always. 

 The growth of negative residuals in the column 0-Y is too obvious 

 and too serious to allow of the idea of a uniform arcual velocity. As 

 remarked in the last Eeport, such a velocity would make Y cross S, 

 preceding it up to about 105° and following it after that. The recoi-ds 

 discussed in the last Eeport were all in the neighbourhood of 95°-100°, 

 where the residuals — Y are seen to be comparatively small; the later 

 ones are inconsistent with the crossing of S. Apparently Y'' (we may 

 perhaps still retain this letter for the phenomenon, whatever it is) 

 always precedes S [and incidentally it may be remarked that this fact 

 really increases the chances of its being mistaken for S and so causing 

 the apparently greater uncertainty in identification of S which is so 

 curious, seeing that on any given record S is better marked than P] . 

 Its time of transmission may be put as follows: — 



A = 95 100 



s. s. 



Y = 1420 1470 



105 



8. 



1490 



110 



8. 



1520 



Is it some combination of P and S ? If we add together the times 

 forP and S as given by the Shide tables so as to obtain these figures 

 we get 



But, of course, as the tables stand, the values of 8P and SS for 

 such arcs are quite unequal, so that no effective combination is possible. 

 If, however, we further modify the curve of 8P shown in fig. 4, so 

 that the max. near 30° is followed by a minimum near 40°, and this 

 again by a maximum near 60°, then possibly we can get the values 

 of 5P and 8S equal. Assuming S to be throughout in the ratio 1-79 

 to P, the values of SP near 40° and 60° must be in this ratio. Thus 

 if SP falls again to 4 at 40°, it must rise to 7 at 60°, which is far from 

 unreasonable. A provisional set of tables has been framed on these 



