859 
The mean error in the estimations of silver amounts to 3 °,,. 
In the measurements attention should be paid to the following points. 
pall, 
: 
{ — 
40 
JE on Ab 8x vergroot. ZB 
wh by 
130 
120 
118 
100 
90 
—S 
80 
70 
60 
30 
30 / I 
/ 2 
20 f af ak 
5 OT le el Oka Za arl 
005 0 05 020 Omg 
A AMER Aven A ee A A RI 
| 2 3 4 
Fig. 4. 
1. It appears experimentally that the area of flaking becomes 
larger by ultra-violet rays. To obviate this complication a dish of 
quinine-bisulphate has been placed in the light-path towards the 
liquid to be measured. 
2. The area of flaking is greatly dependent on the concentration 
of the acid, Therefore the same acid concentration has always 
been used. 
3. The area of flaking increases with the temperature. At 17° 
the area increases by 2,7°/, per degree of temperature increase. 
Therefore the temperature of the liquid is always measured before 
and after the flaking, and all the observations have always been 
reduced to the same temperature. 
4. In consequence of specks of dirt and premature flaking through 
the everywhere present sodium chloride, the liquid which is to be 
examined, will already exhibit a beginning turbidity. This error can 
become pretty great, especially with small concentrations. 
By means of suitable manipulations the extent of this effect can 
be registered separately for every flaking. With only a few excep- 
tions we find that this error is always smaller than 5°/,. In many 
56* 
