s indicated
by arrows, into the down-takes marked E, and out through an
annular flue, where they are passed under a boiler.

"It will be noticed that the ore is exposed to the waste heat
on three sides of the retorts, and owing to the great surface
so exposed, the ore is very thoroughly deoxidized, and reduced
in the retorts before it is introduced into the puddling
furnace for final reduction. The curved cast iron pipes marked
D are provided with slides, and are for the purpose of
introducing the deoxidized ore into the second bottom of the
furnace. As before stated, the furnace is intended to
accommodate two charges of ore, and as fast as it is balled up
and taken out of the working bottom, the charge remaining in
the second bottom is worked up in the place occupied by the
first charge, and a _new_ charge is introduced. As fast as the
ore is drawn out from the retorts the elevator supplies a new
lot, so that the retorts are always filled, thus making the
process continuous."

The temperature of the charge in the deoxidizer is from 800 deg. to 1,000 deg.
F.--_Amer. Engineer._

* * * * *

SOME REMARKS ON THE DETERMINATION OF HARDNESS IN WATERS.

By HERBERT JACKSON.

Having had occasion some short time ago to examine a hard water which
owed half its hardness to salts of magnesium, I noticed that the soap
test, applied in the usual way, gave a result which differed very much
from that obtained by the quantitative estimation of calcium and
magnesium. A perfectly normal lather was obtained when soap had been
added in quantities sufficient to neutralize 14 deg. of hardness, whereas
the water contained salts of calcium and magnesium equivalent, on
Clark's scale, to a hardness of 27 deg..

Although I was aware that similar observations had been made before, I
thought that it might be useful to determine the conditions under
which the soap test could not be depended upon for reliable results.

I found with waters containing calcium or magnesium alone that,
whenever salts of either of these metals were in solution in
quantities sufficient to give 23 deg. of hardness on Clark's scale, no
dependence could be placed upon the results given by the soap test. In
the case of waters containing salts of both calcium and magnesium, I
found that if the salts of the latter metal were in solution in
quantities sufficient to g