1318 
In the nickel-spectrum the pressure-effect is not proportional to 
the increase of the pressure. 
A general study of the question how then the pressure-effect changes 
with the pressure cannot be made from the present observation- 
material. In general there is no regularity to be seen, though for 
most of the lines B, > B, which indicates that for a strong increase 
in pressure the shift changes less, but to this rule there are excep- 
tions e.g. group 1 (4 3524.69 where B, << B,, and group Il (4 3610.68 
and 4 2807.29) where B, =B.,. 
Still it is remarkable that the groups I and Il have the same 
mean value of 4, and B, (2.8 resp. 1.9). For group IV these values 
(181 resp. 8.1) are in a ratio to each other which is a little 
greater than 2.8 and 1.9. Of the groups III and V nothing can be 
said because there are too few observations of B. 
That for a pressure-increase of 5 atmospheres we get a larger 
shift than for one of 10 atmospheres as is the case for 43548,32 
(Group I) (for 5 atm. 5 X 5.0 = 25 and for 10 atm. 10 2.1 = 21) 
will make the explanation of the pressure-effect still more difficult 
A similar case is found for 23972.32 (Group II), for which Birnam 
observed the same shift (28) for 5 and for 10 atmospheres pressure- 
increase, from which it follows, that an increase of the pressure 
from 5 to 10 atmospheres does not shift the spectrum line. 
Let us now consider DurrieLp’s observations. Here also we find, 
that the shift per atmosphere calculated from the lower pressures 
(10 atm.) (D,) is generally larger than that calculated from the 
observations at higher pressures (20—100 atm.) (D,). Exceptions are 
found in group I for 23783.67, which line shows in the are-spectrum 
a shift proportional to the pressure-increase; further for 4 3775.74 
in group Il where at a stronger pressure-increase the shift increases 
more than at a pressure-increase to 10 atm. Something similar, 
though in a less degree, is found in group III for 2 4359.70 and in 
group V for A 4401.77. 
tases in which the differences between the observations at low 
and high pressure are very pronounced are found e.g. for 2 3597.86 
and 23610.68 of group II. 
As to the four first groups the ratios between the means of D, 
and D, are then higher than those of the means of /, and B, as 
is shown in this table. For the lines of group V the ratio between D, 
and JD, for one and the same line approaches more and more to one. 
The comparison between the columns 4, with D, and D, has 
been made by Dourrierp *) already, who has however not extended 
ete: 
