4 BULLETIN 524, U. S. DEPARTMENT OF AGRICULTURE. 
The salt-free ash in duplicate analyses should agree within 0.01 per 
cent, and further analysis should be made if the results do not agree 
within this limit. 
Disagreements in the total ash are considered as due to volatiliza- 
tion of sodium chlorid at different temperatures of ignition and not 
to variation in sampling. If the differences were due to variation in 
sampling the ash of unsalted butter should show the same relative 
discrepancies. To show that this is not the case, ash percentages of 
three unsalted butters are given as follows : 
Table 3. — Agreement in ash percentages of unsalted outters. 
Sample 
No. 1. 
Sample 
No. 2. 
. Sample 
No. 3. 
Per cent. 
0.139 
.141 
.137 
Per cent. 
0.202 
.204 
.205 
Per cent. 
0. 0S8 
.087 
.086 
The differences between triplicate determinations are well within 
the 0.01 per cent limit. 
The amount of sample to be used in making the calcium oxid 
determinations will depend on the capacity of the dishes avail- 
able or the quantity of ash present. In the experiments subse- 
quently reported the ash of from 10 to 50 grams of sample was 
used, treated as follows: A known quantity of decinormal hydro- 
chloric acid was added, warmed to dissolve the ash, and after cool- 
ing the solution was titrated back with decinormal sodium hydroxid 
to obtain the' alkalinity, using Methyl Orange as the indicator. 
Acetic acid was then added in slight excess and the calcium pre- 
cipitated from hot solution with ammonium oxalate. After stand- 
ing overnight, preferably on a warm steam bath, the calcium oxa • 
late was collected on a small filter, washed free from chlorid, and 
ignited to constant weight as calcium oxid. The volume of solu- 
tion when filtered should not be over 100 cc. Under conditions of 
slight acidity and small volume the error in the calcium determina- 
tion, due to the solubility of calcium oxalate in acetic acid solution, is 
negligible. When iron or aluminum is present the ferric or alumi- 
num phosphate, being insoluble in acetic acid, must be filtered off 
before the precipitation with ammonium oxalate. Practically, this 
is necessary only in those cases where cream or butter has become 
contaminated with iron from the container. Leach 1 gives the per- 
centage of iron oxid in the ash of milk as 0.13 and makes no mention 
of aluminum. The average ash of milk is 0.71 per cent, cream has 
less ash, depending on the amount of fat, and the ash of unsalted but- 
1 Leach, Albert E., Food Inspection and Analysis, 3d ed., p. 12S, New York, 1913. 
