(hiAi-. XXXVI.] nagpuk: MAM)ia. TcjO 
Mn, Fe, SiO->, and P, tlie calculated analysis of the sample before 
cleaning works out as follows : — 
Mangaiu'sc 4'.)-7H 
Iron «-77 
Silica «-8'» 
Phosplu.nis 0-097 
Comparison of this with the analysis of samj)le ."51 indicates that the two 
portions into which the original sample was divided were not exactly 
alike, and that either (1) for the amount of ore taken the pieces of ore 
should have been broken to a somewhat smaller size than they were to 
ensure the final sample obtained on reduction being truly representative of 
the whole, or (2) the weight of the original sample should have been consi- 
derably larger, if the size of the pieces of ore was to be kept the same, 
namely 1 to 2 inches in diameter, for the first halving. Considering then 
only the portion of the sample which was subjected to the cleaning opera- 
tion, that is considering only analyses 32 and 33, it is evident that the residt 
of cleaning is to raise the analy.sis of the sample from 49 -78 per cent, ot 
manganese to 51-11 per cent. This improvement is, however, much less 
than I should have expected and shows that a relatively large amount 
of visible spessartite can be present in the ores without the quahty of 
the sample being very seriously lowered, although, when prices rule low, a 
dif?ereiice of H per cent, in the percentage of manganese present may 
make all the diflerence between whether the ore can be exported at a 
profit or not. The conclusion to be drawn from this experiment is that 
when the prices are high a fair amount of spessartite-bearing material may 
be allowed to remain in the ore, but that when the prices are low it may 
be advisable to clean out this foreign material and raise the quality 
even at the cost of a considerable rejection of ore (in this case 40 per 
cent). The result? of this experiment should not, however, be taken as 
generally applicable as regards the amount of improvement to be 
effected by cleaning. It merely indicates the method to be adopted 
in determining the advisability of subjecting the ores of a particular 
mine to rigid cleaning. Each case should be decided on its merits by 
means of an experiment similar to that described here. For convenient 
cleaning the pieces of ore should be not smaller than one inch in diameter. 
That is to say the pieces in the sample before the initial division of it 
into two portions, one to be reduced without cleaning, and the other to be 
cleaned, should not be smaller than the size mentioned above. As the 
size of the pieces of ore must not be too large in proportion to the size of 
the sample to be halved, it would be as well to take about 1,000 lbs. of 
ore for the initial halving, and so avoid the difference obtained in my 
experiment between the analyses of the two halves of the sample, the 
one given by sample 31 and the other calculated from samples 32 and 33. 
