(ATL) 
solutions astonished me somewhat as during the boiling a certain 
amount of mixing takes place by the rising of the vapour bubbles. 
The difference in temperature between two aqueous layers of a boil- 
ing watercolumn at a distance of 1 c¢.m. of each other ought, 
theoretically, to amount to = 0.036°. My actual experiments gave 
values laying between 0.015 and 0.030°. A change in the position 
of the thermometer had a smaller influence, when the instrument was 
deeply immersed in the liquid than when it was: nearer the surface. 
This is only natural if we remember, that the vapour bubbles ascen- 
ding from the bottom layers are larger in size than those of the 
top layers so that the mixing process in the former is less imperfect. 
In any case it was shown that the “mixing” during the boiling 
was by no means sufficient to quite neutralize the difference in tem- 
perature between the different layers of liquid. As however the 
difference in temperature is partly neutralized, we are dealing here 
with a state of labile equilibrium. 
After having tried BerCKMANN's process, I applied the method 
proposed by S. SAKURAI!), because the results, obtained by W. 
LANDSBERGER *), who used this process in a slightly simplified form 
for the determination of molecular weights, inspired me with confi- 
dence. On further investigation however, it appeared to me that no 
very accurate results are obtainable by this method. As was to be 
expected, the boiling point of the water or of the solution depended 
on the temperature of the steam which was blown into it. Unless 
the evaporation keeps perfectly equal pace with the condensation, 
the pressure in the steam-generating flask is lable to constant varia- 
tion and as this affects the temperature of the steam we cannot be 
certain of a constant boiling point. For this reason J abandoned the 
method. 
All this induced me to have a new boiling apparatus constructed 
with the object of avoiding the first two sources of error arising from 
the use of BECKMANN’s apparatus, although convinced that the third 
objection would still remain. 
As we know by experience that a liquid boils more readily in a 
metallic vessel with a rough interior surface than in a glass one, in 
other words that, even without special precautions, the danger of 
overheating is smaller when using a metallic vessel instead of a glass 
one, a metallic vessel seemed to me preferable to one made of glass 
and I, therefore, decided to determine the boiling points in vessels 
1) Journ. of the chemical Society 63. 495. 
*) Zeitschr. f. Anorg. Chem. 17. 423. 
