262 
PHYSICS: G. R. HARRISON 
Proc. N. a. 
intensity of unresolved line absorption and the unabsorbed spaces 
between. 
An important point, which was indicated by all photographs showing 
the continuous absorption, is that this does not extend undiminished to 
the extreme ultraviolet. It decreases rapidly with decreasing wave- 
length ; that is, the vapor is much more transparent fifty Angstroms below 
the head of the series than it is right at the head (2413 A). In fact, the 
region at 2200 A is one of the most transparent in the whole range of 
wave-lengths studied. 
From the analogy to a similar case with X-rays, where continuous ab- 
sorption decreases as the cube of the wave-length, it would be of interest to 
know the function determining the falling off of the absorption in the 
present case. It is expected that in a fuller report on this work to be 
published elsewhere definite data on this can be given. 
In figure 1 a set of curves for sodium is given showing transmission of 
light as ordinates and wave-lengths as abscissas. These are given to show 
the manner in which the absorption changes with vapor density through- 
out the visible and ultraviolet. The curves are to be taken as merely 
qualitatively correct, since the photometric method used was necessarily 
rough due to irregularities in the background. 
With unsaturated vapor the band absorption is tremendously weakened 
in comparison to the line and continuous absorption. It is difficult to get 
pictures which can be compared, since we have no way of determining the 
relative amounts of sodium present in the absorbing columns in the two 
cases. The best available method seems to be to take two pictures which 
show line 2853 (say) absorbed to the same degree in both and use this for 
a parameter. Two such pictures being chosen it was found that the 
band absorption was visible around the first four lines in the case of the 
saturated vapor, and about the D-lines alone, very faintly, in the unsatu- 
rated. Potassium is now being tested for a similar effect. 
This evidence seems to indicate that the bands may be due to loose 
molecular aggregates which are broken up when the mean free path becomes 
greater and the temperature motion more violent. vSodium vapor is 
supposedly monatomic, but Hackspill's^ vapor pressure determinations 
seem to show that loose aggregates may exist, since the results with satu- 
rated vapor are not very concordant. 
It was noticed that the fluorescence, which was quite marked in saturated 
vapor, practically disappeared on superheating. Van der Lingen and 
Wood^ found a similar effect in mercury vapor. The intimate connection 
between the mechanisms of fluorescence and band absorption has been 
definitely established by the work of Wood and others. 
Results with Potassium. — This metal showed similar continuous ab- 
sorption to that of sodium, the saturated vapor only having been studied 
