Fune 24, 1886] 
plan of experiment chosen by Prof. Newcomb, and 
that pursued by Prof. Michelson in his similar in- 
vestigation at the Naval Academy in 1879 (see NATURE, 
vol. xxi. pp. 94, 120). Fig. 1 represents in principle the 
arrangement adopted by the former, which was also that 
used by Foucault. In it the lens, 1, is placed between 
the light-source, S, and the revolving mirror, A. Fig. 2 
shows the disposition preferred by Michelson, in which 
the lens is interposed between the revolving and fixed 
mirrors. In both equally, S and M are, and for the 
purpose in view necessarily must be, in conjugate foci of 
the lens. 
A disadvantage of the first form is that the measure- 
ment of any considerable deviations will be attended by 
uncertainties caused by the oblique passage through the 
lens of the return beams. It was, however, obviated in 
the experiments under consideration, by the use of Zwo 
lenses—one for the outgoing, the other for the incoming 
rays. The second method (Michelson’s) promises in- 
creased brilliancy of the image ; which may, nevertheless, 
be regarded as outweighed by atmospheric and other im- 
A 
Fig.l. 
pediments to its distinctness, as well as by the illumination 
of the field of view produced by the passage through it of 
some part of the lens with every revolution of the mirror. 
The method exemplified in Fig. 1 was then chosen by 
Prof. Newcomb as affording more or less calculable 
conditions ; while No. 2 involved all the uncertainties of 
definition habitually besetting astronomical observations. 
Let us now endeavour to realise the nature of the ex- 
perimenter’s immediate task. The precise measurement 
of an angle actually constitutes it. From the mirror A, 
so long as it remains at rest, an image is reflected in a 
certain direction ; but no sooner is A set rapidly rotating, 
than the same image is reflected in a slightly different 
direction. The amount of this difference—in other 
words, the angle of deviation—is the object to be 
ascertained. 
Obviously, the first desideratum is to render the inevit- 
able error of measurement comparatively small, by making 
the quantity to be measured large. Two roads are open 
towards this end. A high velocity can be given to the 
murror A; or a great distance can be interposed between 
Aand M. By the first means, the angle rotated through 
an a given time will be augmented ; by the second, the 
NARORE 
171 
time available for the displacement of the reflector will be 
prolonged by the lengthening of the journey imposed 
upon the rays to be reflected. The difficulties hampering 
increased speed are purely mechanical, though none the 
less formidable ; those in the way of a lengthened path 
are optical. 
The preservation of light enough to keep the image 
bright and distinct is of paramount necessity for the 
avoidance of ruinous uncertainties in its measurement. 
Now, in Foucault’s experiments, the object affording the 
image was the line of a reticule. It was dark upon a 
bright ground ; a platinum-wire relieved against a sheaf 
of sunbeams. But no perfectly defined image of such an 
object could be formed at any considerable distance ; and 
we find, accordingly, that the utmost length by which he 
ventured to separate his mirrors was twenty metres. His 
entire apparatus was, in fact, contained in a single room. 
Hence, notwithstanding a speed given to his mirror of 
from 600 to 800 revolutions per second, the actual linear 
deflection of the return ray amounted to no more than 
seven-tenths of a millimetre. Chiefly by employing as 
om 
Te 
(Sara 
A 
Fig.2. 
his light-source an illuminated slit, the lucent image of 
which on a dark ground bore the enormous loss of light 
ensuing from the transportation of the fixed mirror to a 
distance of close upon 2000 metres, Michelson was 
enabled to augment this deflection some two-hundred- 
fold. The resulting velocity for light of 299,910 kilometres 
per second was proportionately trustworthy, the error of 
the angular measurement upon which it immediately de- 
pended being estimated to be one hundred times less 
than in Foucault’s determination. Prof. Neweomb’s im- 
provements carried him still further towards absolute 
accuracy. 
The details of construction of his “ phototachometer ” 
were decided on in the summer of 1879, and the instru- 
ment was completed by the Messrs. Clark in May 1880. 
It consisted essentially of four parts—a sending and a 
receiving telescope, a revolving and a fixed mirror. Sun- 
light, thrown from a heliostat through an adjustable ver- 
tical slit at the eye-end of the sender, passed down the 
tube, which was bent at right-angles to get it out of the 
way of the observing telescope, and after reflection by a 
plane mirror at the elbow, passed out through the objective 
towards the revolving mirror. This was formed by a 
