contained some iron. The first cake, No. 7, gave off during ignition an
agreeable odor resembling some of the finer tobaccos, and this is
characteristic more or less of all the cakes. The ash weighed 52 per
cent., the soluble part of which, 18.5, was mostly potassium carbonate,
with some chlorides and sulphates; the insoluble, mostly chalk with iron
and alumina. No. 8--highest priced of all--had in the mass an odor which
I can compare to nothing else than a well rotted farmyard manure. Twenty
parts of the ash were soluble and largely potassium carbonate, the
insoluble being iron for the most part. The mineral portions of Nos. 9
and 10 closely resemble No. 7.

On looking over the results, it is found that the red rolls contained
starchy matters in abundance (in No. 4 the starch was to a large extent
replaced by water), and an ash, mostly sodium chloride, introduced no
doubt to assist in its preservation as well as to increase the color of
the resin--a well known action of salt on vegetable reds. The cakes,
which are mostly used for cheese coloring, I believe, all appeared to
contain turmeric, for they gave a more or less distinct reaction with
the boric acid test, and all except No. 8 contained large quantities of
chalk. These results in reference to extractive, etc., reveal nothing
that has not been known before. Wynter Blyth, who gives the only
analyses of annatto I have been able to find, states that the
composition of a fair commercial sample (which I take to mean the raw
article) examined by him was as follows: water, 24.2; resin, 28.8; ash,
22.5; and extractive, 24.5; and that of an adulterated (which I take to
mean a manufactured) article, water, 13.4; resin, 11.0; ash (iron,
silica, chalk, alumina, and common salt), 48.3; and extractive. 27.3. If
this be correct, it appears that the articles at present in the market,
or at least those which have come in my way, have been wretched
imitations of the genuine thing, and should, instead of being called