NEW INSTRUMENT FOR GRAVITY MEASUREMENTS 25 
at two successive stations. Any hysteresis effect is then only a small fraction 
of this amount. This sounds simple but it is difficult to carry out for it re- 
quires that the suspended mass be locked to within less than one tousandth 
of a millimeter of the true position. To meet this requirement and to avoid 
any strain on any part of the spring from jolts in transportation has neces- 
sitated a very elaborated locking mechanism. 
In connection with this is another matter that no one seems to think of 
unless he has been working with a seismograph; that is vibration. The ground 
is constantly vibrating, and a weight hung on a spring is an excellent seismo- 
graph if its motions can be closely observed, so some way must be found to 
reduce the sensitivity to vibrations while maintaining the sensitivity to 
changes in the static force. In the present instrument this is accomplished to 
a certain extent by excessive air damping but this is only satisfactory under 
very favorable conditions so some modifications are being planned for the 
beam system whicn will give a more favorable ratio between these two effects. 
In the city traffic vibration is the chief difficulty and in the country it is. the 
wind in the trees, so it is impossible to use the instrument in very windy 
weather but with a light wind readings can be made by watching for a favor- 
able mement, the mirrors may be almost stationary for several seconds at a 
time and an experienced operator can utilize these moments to secure read- 
ings to about one part in a million. 
One more difficulty encountered in the construction of the instrument 
was due to initial strains in the frame. Any change in temperature would 
alter the elastic forces present and cause the frame to twist enough to throw 
the beam system badly out of adjustment. To avoid this as far as possible the 
beam and mirror system is carried by a separate frame entirely independent 
of that which carries the rest of the mechanism. This frame is made of an- 
nealed aluminum tubes very accurately fitted into sockets in the top and 
bottom plates, and clamped in position by a method designed to avoid put- 
ting any strain in the tubes when they are tightened up. 
It was found, in the first experimental instrument, that the temperature 
stresses together with stresses due to imperfect construction of the locking - 
mechanism would pull the beam and mirrors slightly out of alignment when 
it was locked. This caused distortion of the filament supports of the mirrors 
and of a light guide spring that was used in that instrument and when re- 
- leased the filaments would not immediately return to their relaxed condition 
but there would be a slow drift of the readings extending over a period of 
from half an hour up to several hours before the correct reading was obtained. 
In the latest instrument this effect is almost eliminated but it is sometimes 
necessary to wait about ten minutes. 
These effects due to elastic hysteresis have been the most serious diffi- 
culties encountered, and there is still room for improvement in this direction, 
but measurements are now being made than can be checked, day after day, 
to +1-10-* and, as soon as a small alteration in the mirror system can be 
completed, which will give a better ratio between the effect of gravity and the 
effect of vibrations, I expect to get an accuracy four or five times as great. 
169 
