making a vertical scale of grays from black to white
(Chapter II., paragraph 25).

If left to personal preference, an estimate of middle value will vary
with each individual who attempts to make it. This appears in the
neutral scales already published for schools, and students who depend
upon them, discover a variation of over 10 per cent. in the selection of
middle gray. Since this VALUE SCALE underlies all color work, it needs
accurate adjustment by scientific means, as in scales of sound, of
length, of weight, or of temperature.

A PHOTOMETER (_photo_, light, and _meter_, a measure)[19] is shown on
the next page. It measures the relative amount of light which the eye
receives from any source, and so enables us to make a scale with any
number of regular steps. The principle on which it acts is very simple.

[Footnote 19: Adopted in Course on Optical Measurements at the
Massachusetts Institute of Technology. Instruments have also
been made for the Harvard Medical School, the Treasury
Department in Washington, and various private laboratories.]

A rectangular box, divided by a central partition into halves, has
symmetrical openings in the front walls, which permit the light to reach
two white fields placed upon the back walls. If one looks in through the
observation tube, both halves are seen to be exactly alike, and the
white fields equally illuminated. A valve is then fitted to one of the
front openings, so that the light in that half of the photometer may be
gradually diminished. Its white field is thus darkened by measured
degrees, and becomes black when all light is excluded by the closed
valve. While this darkening process goes on in one-half of the
instrument, the white field in the other half does not change, and,
looking into the eyepiece, the observer sees each step contrasted with
the original white. One-half is thus said to be _variable_ because of
its valve, and the other side is said to be _fixed_. A dial connected
with the valve has a hand moving over it to show how much light is
admitted to the field in the variable half.

Let us now test one of these personal decisions about middle value.
A sample replaces the white field in the fixed half, and by means of the
valve, the white field in the variable half is alternately darkened and
lightened, until it matches the sample and the eye sees no difference in
the two. The dial then discloses the fact that this supposedly MIDDLE
VALUE refle