32 
to that produced by the addition of the same reagent to 
70 cc. of distilled water to which 1 cc. of the standard has 
been added, v T e shall have 1 grain per gallon of lead in the 
water. But one-tenth of a grain of lead per gallon is generally 
considered hurtful when present in a drinking water; to 
estimate this we should require to use only 0T cc. of the 
standard : a very small error in reading the burette measure- 
ments would introduce a comparatively large error in the 
result. Thus in the case of a water containing one-tenth 
grain of lead per gallon an error in reading of 0'05 cc. would 
introduce an error in the quantity of lead equal to one-half 
of the total quantity to be estimated. The first point, there- 
fore, to investigate appeared to be the strength of the 
standard solutions. I shall describe the experiments made 
with copper. 
Standard used 1 cc. = 1 mgm. copper. 
Expt. Taken. Found. 
No. 1 5 mgm. per litre... 4 mgm. per litre. 
No. 2 2-5 ,, „ 2 „ „ 
Standard used 1 cc.r^OT mgm. copper. 
Taken. Found. 
No, 3 5 mgm. per litre... 4 ’8 mgm. per litre. 
No. 4 2-5 „ „ 2-4 „ 
No. 5 0-5 „ „ 0-5 „ 
In each case 50 cc. of liquid was used. 
Similar results were obtained with lead solutions. 
The use of a standard, 1 cc. of which is equal to 0T mgm., 
of copper or of lead enables more accurate and more delicate 
results to be obtained than the use of a stronger standard 
does. 
The second point to be determined was the limits of 
accuracy of the method, and first as to the lower limit. 
From the experiments with copper already detailed it will 
be seen that 0*5 mgm. of copper per litre could be estimated 
by using 50 cc. of the liquid. 
