502 
THE EARL OF BERKELEY AND MR, E. G. J. HARTLEY ON THE 
Table IX. 
I. 
II. 
III. 
u 
HH 
Y. 
VI. 
VII. 
IX. 
X. 
Date. 
Name 
of 
tube. 
Apparent 
turning- 
point 
pressure in 
atmo¬ 
spheres. 
Rate of 
movement 
of water- 
level due 
to guard¬ 
ring leak. 
Turning- 
point 
pressure in 
atmospheres. 
Time. 
Solution 
leak. 
The 
increment 
of pressure 
used in 
the experi¬ 
ment. 
Rate of 
movement 
of water- 
level 
caused by 
increment 
of 
pressure. 
h. m. 
atmospheres. 
250 
Grammes Galactose. 
Feb. 2 
X 
35-58 
0-7 
*35-47 
3 35 
0-0002 
0-34 
2-8 
380 
Grammes Galactose. 
March 14 
X 
62-86 | 
0-6 
*62-79 
2 35 
none 
0-34 
2-9 
500 
Grammes Galactose. 
March 12 
N 
— 
0-3 
94.59 
3 10 
0-0006 
0-34 
_ 
„ 26 
N 
— 
0-3 
96-63 
2 12 
0-0004 
0-34 
_ 
May 3 
N 
— 
0-4 
*97-31 
4 15 
j.v.t. 
0-34 
— 
„ 11 
N 
94-92 
0-7 
*94-75 
2 30 
0-0003 
0-34 
1-5 
Remarks. — The experiments of March 26 and May 3 were with the recovered 
galactose mentioned on page 497. 
No higher osmotic pressures were determined, as it was found that a solution of 
500 grammes in the litre was already supersaturated at 0° C. 
The Experimental Errors . # 
Unfortunately we are not in a position to evaluate the experimental errors. The 
approximate constancy of the guard-ring leak, with different tubes, seems to justify 
the assumption that no great variable error enters through this cause, but it should 
be pointed out that we have no means of proving that there is not a constant error 
introduced thereby. The errors due to “ solution leak ” are mentioned in tlie next 
section, and are obviated by discarding those experiments which showed a leak ; in 
this connection it should lie remembered that analyses of very small quantities of 
sugar are far from reliable, and it may be that even the results given by those 
experiments which showed no trace of sugar are really too low. 
It was feared that the use of a somewhat more viscous oil than the dead-weight 
apparatus is designed for (the makers recommend glycerine) might in cold weather 
vitiate the results ; special experiments, however, showed this cause of error to be 
absent. It is scarcely necessary to point out that a sensibility of 0T2 atmosphere 
in the' dead-weight gauge, taken in conjunction with a “ guard-ring leak ” rate 
* See Appendix B. 
