172 
NATURE 
[Fune 24, 1886 
rectangular prism of polished steel, 85 millimetres in 
height, and with a cross-section of 37°5 square millimetres. 
The vertical faces constituting the reflecting area were 
nickel-plated, and proved of a remarkably durable though 
not very high polish. Motion in opposite directions at will 
was communicated by two air-turbines, acting one at the 
top, the other at the bottom of the mirror, and serving, by a 
simple contrivance, each for the regulation of the contrary 
velocity imparted by the other. A wheel-work arrange- 
ment, by which an electric current was broken once for 
every twenty-eight revolutions of the mirror, gave the 
means of @btaining a chronographic record of its rate of 
going. Two fixed mirrors, mounted side by side on cast- 
iron stands, were employed to return the light sent to 
them by the revolving mirror. Each was about 40 
centimetres in diameter, and had a radius of curvature of 
some 3000 metres. The object-glass of the receiving 
telescope was (in the first instance) placed immediately 
under that of the sender, the former thus directly facing 
the lower, the latter the upper section of the movable 
mirror. The two /zdes, however, owing to the “ broken” 
form given, as already mentioned, to that of the sender, 
made with each other an angle of go’. Horizontal movement 
round a vertical axis coincident with that of the rotating 
mirror, was possessed by the observing telescope, to which 
was attached a pair of microscopes for reading off the 
divisions‘on a horizontal divided arc fixed to a stiff frame 
at its further end. The amount of this horizontal motion 
of the telescope measured the deviation of the thrice- 
reflected sunbeam, and, by an immediate deduction, its 
velocity. 
The site chosen for the erection of the apparatus was 
Fort Myer, on the south side of the Potomac, overlooking 
the city of Washington. The stationary mirrors, to and 
from which the carefully guarded rays performed their 
trips, were placed, ‘to begin with, in the grounds of the 
Naval Observatory, at a distance of 2551 metres from | 
Fort Myer; but were in 1881 removed to a point at the 
base of the Washington Monument, at a distance increased 
to 3721 metres. Some tentative experiments were under- 
taken on June 22,1880; after a few days’ trial, however, it 
was found that the wheel-work for counting the revolutions 
of the mirror was destroyed by the rapidity of the im- 
pressed movements. New wheels wore out almost before 
a set of readings could be obtained with them; until at 
length the Messrs. Clark, finding that no metal would 
stand the inflictel wear and tear, substituted vaw hide as 
the material ‘for the first wheel, a device which proved 
wholly successful. With the instrument thus modified 
work was begun on August 9, and continued without inter- 
ruption until September 20. The transportation of the 
fixed mirrors to the Monument station in the spring of 
1881 postponed the commencement of operations to 
August 8; and their effective prosecution was then im- 
peded by the discovery of a source of systematic error in 
a “torsional vibration” of the rotating mirror. That is 
to say,-the steel prism employed to reflect the light, no 
longer, when its speed attained a certain point, revolved 
as an absolutely rigid whole, but fended towards the pos- 
session of different velocities in its different parts. Hence 
a slight twisting of its mass producing vibrations round 
the axis of rotation, the effect of which was visible in 
the breaking up of :the image of the slit into four separate 
images, one due to each of the faces of the prism. The 
persistence of this baffling symptom compelled a modi- 
fication of the instrument, by which the sending and 
receiving telescopes could be respectively depressed and 
raised so as to alternate their positions, and the portions 
of the mirror they were directed towards. The mean of 
any two complete sets of observations made with the 
telescopes thus interchanged would be free, as Prof. 
Newcomb shows, from the effects of any probable form 
of torsional vibration. 
No such effects, however, were apparent in the obser- 
vations of 1882. This last series extended from July 24 
to September 5, and were so nearly free from accidental 
differences that the probable error of a complete deter- 
mination was scarcely more, under good conditions, than 
the ten-thousandth part of the whole. Upon these, ac- 
cordingly, the chief reliance was placed in the final 
discussion of results. 
The announcement that Messrs. Forbes and Young 
had detected a difference of 2 per cent. in the rates of 
transmission of red and blue light prompted, at Fort 
Myer, a most careful watch for traces of colour in the 
reflected image of the slit. But although, from a dis- 
crepancy of even one-twentieth that amount, a spectrum 
15” in breadth must have ensued, the iridescent edges 
which would infallibly have betrayed its presence were 
not seen. 
An important novelty in Prof. Newcomb’s method was 
his use of opposite rotations and their accompanying 
opposite deviations. In his instrument the mirror, as 
already stated, could be made to revolve at pleasure, either 
from right to left or from left to right. Instead, then, of 
measuring, as had always previously been done, the de- 
flection produced in the return ray by the change from 
rest to an ascertained rate of rotation, the object of his 
determinations was the total deflection due to extremes 
/ M 
Ep 
in) 
of contrary movement. The mode of experimenting was 
briefly as follows. : 
First, the valve was opened to the air-blast giving 
negative rotation, the receiving telescope being set upon 
some division near one extremity of its arc; the image of 
the slit was then accurately fixed, by the regulating agency 
upon the velocity of the mirror of the opposing air- 
current, upon the middle wire of the micrometer ; the 
chronograph made its record of the rate of going, and 
the microscopes were read. This constituted what was 
called a “run,” and occupied two minutes or upwards. 
The telescope was next unclamped, and directed near the 
opposite end of the divided are. Posztive motion was 
given by opening the other valve, and the process of 
fixing the image and reading off repeated. A comparison 
of the two sets sufficed to determine the time spent by 
the light in passing to and from the mirrors on the other 
shore of the Potomac. : 
This method of contrary deviations is most strongly 
recommended by Prof. Newcomb to future investigators. 
It combines the two advantages of doubling the angle to 
be measured, and of abolishing possible errors in the 
determination of the zero-point. In the present series, 
velocities, alternately in opposite directions, rarely ex- 
