The construction that I have chosen for the scale is represented in fig. 1. (See Plate VI.) It is composed of small wires, about of an inch in diameter, placed side by side, so as to form a scale of equal parts, which may with ease be counted by means of a certain regular variation of the lengths of the wires.

The external appearance of the whole instrument is that of a common telescope, consisting of three tubes. The scale occupies the place of the object glass, and the little lens is situated at the smaller end, with a pair of plain glasses sliding before it, between which the subject of examination is to be included. This part of the apparatus is shewn separately in fig. 3. It has a projection at a, with a perforation through which a pin is inserted to connect it with a screw represented at b, fig. 2. This screw gives lateral motion to the object, so as to make it correspond with any particular part of the scale. The lens has also a small motion of adjustment by means of the cap c, fig. 2, which renders the view of the magnified object dis

tinct.

Before the instrument is completed, it is necessary to determine with precision the indications of the scale, which must be different according to the distance to which the tube is drawn out. In my instrument, one division of the scale corresponds to Too of an inch when it is at the distance of 16,6 inches from the lens; and since the apparent magnitude in small angles varies in the simple inverse ratio of this distance, each division of the same scale will correspond to at the distance of 8 inches, and the intermediate fractions ‰, 7000, &c. are found by intervals of 1,66 inch marked on the outside of the tube. The basis on which these indications were founded in this instrument, was a wire carefully ascertained

1

to be of an inch in diameter, the magnified image of which occupied fifty divisions of the scale, when it was at the distance of 16,6 inches, and hence one division

I

50 X 200

I

10000

Since any error in the original estimate of this wire must per-
vade all subsequent measures derived from it, the substance
employed was pure gold drawn till fifty-two inches in length
weighed exactly five grains. If we assume the specific gravity
of gold to be 19,36, a cylindrical inch will weigh 3837 grains,
and we may thence infer the diameter of such a wire to be
of an inch, more nearly than can be ascertained by any other
method. For the sake of rendering the scale more accurate,
a similar method was in fact pursued with several gold wires,
of different sizes, weighed with equal care; and the subdivi-
sions of the exterior scale were made to correspond with the
average of their indications.

In making use of this micrometer for taking the measure
of any object, it would be sufficient at any one accidental posi-
tion of the tube to note the number on the outside as deno-
minator, and to observe the number of divisions and decimal
parts which the subject of examination occupies, on the interior
scale, as numerator of a fraction expressing its dimensions in
proportional parts of an inch; but it is preferable to obtain
an integer as numerator, by sliding the tube inward or out-
ward, till the image of the wire is seen to correspond with
some exact number of divisions, not only for the sake of
greater simplicity in the arithmetical computation, but because
we can by the eye judge more correctly of actual coincidence,
than of the comparative magnitudes of adjacent intervals.
The smallest quantity, which the graduations of this instru-

98009

ment profess to measure, is less than the eye can really appreciate in sliding the tube inward or outward. If, for instance, the object measured be really, it may appear To or‚‰, in which case the doubt amounts to part of the whole quantity. But the difference is here exceedingly small in comparison to the extreme division of other instruments where the nominal extent of its power is the same. A micrometer with a divided eye-glass may profess to measure as far as of an inch; but the next division is or 50; and, though the eye may be able to distinguish that the truth lies between the two, it receives no assistance within part of the larger measure.

5000