282 SECTIONAL TRANSACTIONS.—A*, Aft. 
coefficient in the simple probability distribution of the particle. 'The values 
of the masses of the proton and electron are determined by this substitution 
of a simple probability distribution for a double probability distribution. 
DEPARTMENT OF COSMICAL PHYSICS (Af). 
(Sir Frank Dyson, F.R.S., in the chair.) 
Friday, September 7. 
Prof. E. A. Mitne, M.B.E., F.R.S.—A popular account of the significance 
of absorption lines in stellar spectra (11.0). 
Dr. T. Dunnam.—The new Condé spectrograph of the Mount Wilson 
Observatory (11.25). 
Prof. O. Struve.— Spectrophotometric investigations at the Yerkes 
Observatory (11.50). 
Prof. J. A. CarroLit.—Accuracy of measurement in spectrophotometry 
(iZs15): 
(1) Instrumental.—The effect of finite resolving power, etc., due to all 
causes may be represented as spreading a monochromatic source into 
a spectrum of intensity distribution K(T), T being the ‘ reduced’ wave- 
length. Thus a true distribution J(T) is observed as O(T) where 
+00 
O(T) = | I(T + t)K(ddt. 
The practical solution and use of this equation is discussed, by the aid of 
Fourier transformation theory. 
(2) Photographic—(a) The ability of a photographic plate to detect 
small changes in intensity distribution over a given region on the plate is 
discussed, and a quantity termed the ‘ Discriminating Power ’ of the plate 
is defined and shown to be a useful criterion, analogous to Resolving Power 
in optical theory, whereby the performance of the plate used under specified 
conditions may be calculated. 
(b) Certain irregularities on a scale large compared to plate grain size 
are noticed and discussed in connection with variation of film thickness, 
measured optically by interference methods. 
(c) An estimate of limiting accuracy under optimum conditions. 
Mr. E. G. WiLL1ams.—Spectroscopic differences between giant and dwarf 
early type stars (12.40). 
The present procedure for determining absolute magnitudes by the 
spectroscopic method is unsatisfactory for the B and O type stars. It 
requires considerable modification and, until this has been effected, the 
intensity of the interstellar line of calcium is as good a criterion of distance 
as any, provided this intensity is measured spectrophotometrically. 
A number of typical early type stars has been selected for study. Their 
spectra show sharp lines and are free from such disturbing influences as 
axial rotation and the presence of emission lines. The stars have been 
