30 EXPERIMENTS WITH THE DISPLACEMENT INTERFEROMETER. 



of September and beginning of October shows an a effect in the curve following 

 c quite inferior to the effect later between g and h, when the changes of tem- 

 perature are much less marked. Moreover, the minimum between / and g is 

 much too large to be associated with the mean temperature minimum, and in 

 fact they do not coincide. The determined rise of the curve at g began much 

 before any corresponding temperature change. The gaps at d and e introduce 

 uncertainties, but nevertheless one would have expected an a minimum there. 

 It seems probable, therefore, that temperature does not act directly on the pen- 

 dulum (expansion of its parts) but acts on it through another system, which 

 is probably the house itself. The marked effect of steam heat is to thoroughly 

 dry out the basement room. It may be inferred that this is accompanied by 

 redistributions of the stresses of the building and that the concrete subfloor 

 responds to the alteration of load. It should be noted that the last curve CC 

 has been dropped 7 seconds to accommodate it in the drawing and that there- 

 fore the a curve CC as a whole lies much above the original a. curve A A, 

 although the average temperatures are not very different. It seems improb- 

 able, therefore, since the curve has much more than recovered, in fact has 

 considerably exceeded its original reading, at about the same temperature, 

 that there can here be any viscous yielding in the apparatus itself. Further 

 consideration will be given in the next section in connection with the steel 

 pendulum. 



Finally, since the inevitable variations here recorded are within 16 seconds 

 of arc, it was out of the question to attempt to attach the interferometer appa- 

 ratus, adapted for reading within hundredths of a second. The work was there- 

 fore abandoned. 



PART II. AN APPLICATION OF THE DISPLACEMENT INTERFEROMETER 



TO THE HORIZONTAL PENDULUM. 



18. Introductory. The displacement of ellipses or of interference lines in 

 the spectrum is probably capable of being photographed for continuous reg- 

 istry, though less easily than the motion of a spot of light. At all events, it 

 seemed interesting to endeavor to register the excursions of the horizontal 

 pendulum by displacement interferometry, not so much with a view to re- 

 cording seismological phenomena, as to approach by this means certain 

 other problems, the tilting of the earth's surface relatively to the plumb-line, 

 the measurement of the constant of gravitation, etc. The present paper, there- 

 fore, undertakes a new departure with this special end in view, with possibly 

 some ulterior bearing on microseismology. 



If the inclination of the axis of the horizontal pendulum is but a few degrees 

 to the vertical and a large framework is in question (there is scarcely any limit 

 to size other than strength of the material), the sensitiveness of the apparatus, 

 when the excursions are read off in terms of light-waves, is astonishing; or at 

 least it would be so if the instrument supplied with mirror and screen had not 

 been so thoroughly perfected. The horizontal pendulum, moreover, has this 



