372 
DR. T. R„ MERTON AND MR. S. BARRATT 
(6), 40, p. 240, 1920 ; ibid. (6), 41, p. 566, 1921), and Vegard (ibid. (6), 41, p. 558, 1921), 
but it would appear doubtful whether any conclusive evidence as to the origin of the 
spectrum can be obtained by these methods. 
In their investigations of the widths of spectrum lines Buisson and Fabry (' Jonrn. 
de Phys.’ (2), p. 442, 1912) were led to conclude that the secondary spectrum was to 
be referred to the Hydrogen atom. This conclusion was based on a measurement of 
the limiting order at which interference fringes could be observed for a line in the 
secondary spectrum, the relation between the limiting order of interference N and the 
mass M of the radiating particle (in terms of the Hydrogen atom as unity) being given 
by the expression N = k x /(M/6), where k is a constant and 0 the absolute temperature 
(cf. Bayleigii, £ Phil. Mag.’ (6), vol. 29, p. 274, 1915 ; Schonrock, ‘ Ann. der Phys.,’ 
20, p. 995, 1906). 
For the constant k Buisson and Fabry, following Schonrock, adopted the value 
1-22 X 10 e , and the value of N which they found experimentally was in approximate 
agreement with the view that the spectrum was to be referred to the atom. This result 
has recently been criticised by Saha ( £ Phil. Mag.’ (6), 40, p. 159, 1920) on the ground 
that Buisson and Fabry obtained a much smaller value for N in the case of the line 
Ha, the first member of the Balrner series, and that if the value of k in the formula given 
above were calculated from the observed limit of interference for the line Ha, the 
secondary line would yield a value of M more nearly appropriate to the molecule H.. 
The ground of this criticism does not appear to us to be justified, for it is well known that 
the line Ha is complex and could not therefore be expected to yield results in accord¬ 
ance with the theory for a single line, and moreover Saha appears to have overlooked 
the fact that the value of k adopted by Buisson and Fabry was tested experimentally 
with lines of the rare gases, and was found to give results in close agreement with the 
known atomic weights of these gases. 
It must, however, be pointed out that any cause of broadening of the lines other than 
that due to motion in the line of sight will yield too low a value for the mass of the 
radiating particles ; and it follows that measurements of this kind can only set an 
inferior limit to the mass, unless the possibility of broadening of the lines bv any other 
cause can be excluded. 
In the present investigation we have remeasured the wave-lengths of the lines of the 
secondary spectrum in International Angstrom units, and have been able to add a con¬ 
siderable number of lines to those hitherto recorded. We have also investigated the 
effect of variations in the conditions of electrical excitation, and of the pressure in the 
discharge-tube, on the relative intensities of the lines, and have compared the results 
obtained with other methods of classification ; previous investigations (Merton, loc. cit.) 
of the changes in the spectrum which are produced by the admixture of Helium have 
been extended to the more refrangible regions. The widths of several lines in the 
'secondary spectrum have been measured by a new method with which it is believed 
that a high degree of precision has been attained, and under conditions of excitation 
