ter as a natural test plate.

Since the above was written the following details of his exact course
of procedure have been sent to me by Mr. Brashear, and I hereby tender
my thanks:-

"It really takes years to know just what to do when you reach that
point where another touch either gives you the most perfect results
attainable, or ruins the work you have already done. It has taken us
a long time to find out how to make a flat surface, and when we were
called upon to make the twenty-eight plane and parallel surfaces for
the investigation of the value of the metre of the international
standard, every one of which required an accuracy of one-twentieth of
a wave length, we had a difficult task to perform. However, it was
found that every surface had the desired accuracy, and some of them
went far beyond it.

It is an advantage in making flat surfaces to make more than one at a
time; it is better to make at least three, and in fact we always grind
and 'fine' three together. In making speculum plates we get up ten or
twelve at once on the lead lap. These speculum plates we can test as
we go on by means of our test plane until we get them nearly flat. In
polishing them we first make quite a hard polisher, forming it on a
large test plane that is very nearly correct. We then polish a while
on one surface and test it, then on a second and test it, and after a
while we accumulate plates that are slightly concave and slightly
convex. By working upon these alternately with the same polisher, we
finally get our polisher into such shape that it approximates more and
more to a flat surface, and with extreme care and slow procedure we
finally attain the results desired.

All our flats are polished on a machine which has but little virtue in
itself, unmixed with brains. Any machine giving a straight
diametrical stroke will answer the purpose. The glass should be
mounted so as to be perfectly free to move in every direction--that
is to say, perfectly unconstrained. We mount all our flats on a piece
of body Brussels carpet, so that every individual part of the woof
acts as a yielding spring. The flats are held in place by wooden
clamps at the edges, which never touch, but allow the bits of glass or
metal to move slowly around if they are circular; if they are
rectangular we allow them to tumble about as they please within the
frame holding them.

For making speculum metal plates either plane or concave we use
polishers s