L675 
The composition of the mixture was found to be 61°), O,, 36°/, N,, 
SEAN 
A mixture containing less oxygen than ordinary liquid air was 
obtained by collecting the products of combustion of an air-hydro- 
gen flame. This mixture was condensed, the unburnt hydrogen 
escaping in the process, and yielded the following result: 
wt Ol = O—0,03797 dh od. = 000490 TE ail, 020" ju — 0,841 
7 = 0,001599. 
The composition was found to be 9,5°/, O, , 89°/, N, ,1,5°/, Ar’). 
Finally a set of observations were taken with nearly pure oxygen 
obtained from the “Nederlandsche Maatschappij voor zuurstof- en 
stikstof bereiding” at Velsen. These gave as results: 
“= 8962, d= 0,046833 , d, + d, = 0,00470, T= 21,035 „uw 1,148 
n= 0,001895, 
the composition being 96°/, O, , 1°/, N, , 3°/, Ar’). 
The results communicated here regarding mixtures of oxygen and 
nitrogen can only be looked upon as preliminary ones on account of 
the admixture of argon; it is our intention to repeat the observations 
with pure substances. For this reason we do not wish to discuss our 
results in the mean time. 
1) In this case the proportion of Argon was determined by taking the ratio 
to Nitrogen the same as in air. 
2) The proportion of oxygen given here is the result of an analysis of the gas 
before condensation; the proportion in the liquid may be assumed not to have 
been much higher. 
Starting from this oxygen by admixture of liquid air mixtures poorer in oxygen 
were prepared, the densities of which were measured as before. These measure- 
ments agreed well with the density of pure oxygen at the boiling point as deter- 
mined in the Physical Laboratory of Leiden (r = 90°,12, u = 1,147; liquid argon 
having about the same density as oxygen, this figure does not change much by 
a small admixture of this substance) and were found to be properly represented 
up to 84° by the relation : 
u= 1,146 + 0,008 (r — 90) — 0,0015 (r — 90)? — 0,00002 (r — 90)? 
The statement made by BEHN and Kiesrrz (l. c.) that in the neighbourhood 
of pure oxygen the density does not change appreciably with the boiling point, is 
thereby proved to be incorrect, 
