Density of the /Ether. 497 



Inscribed square, the two half beams going in opposite 

 directions ; they could then be re-united by the same semi- 

 transparent glass, and the interference bands observed in a 

 telescope with a micrometer eyepiece. 



During February 1893 a large number of experiments were 

 made in this way. Sometimes the light was sent five or six 

 times round, when only air was in the core, so that no cover- 

 glasses were required ; sometimes water was inserted in the 

 core, and then the light could be got three times round. 



19. Sometimes, in the early trials, with a liquid in the 

 cores, a darkening was seen when the current was put on, but 

 never a shift of the bands ; the darkening was capricious 

 and could not always be got : it was suspected to be 

 occasionally due to a slight shift of the bobbins through 

 mutual magnetic attraction. Polarized light too w^as often 

 employed ; it emerged still fairly plane-polarized, if the plane 

 of polarization was parallel or perpendicular to the plane of 

 reflexion at the mirrors ; otherwise of course it became 

 elliptical. But still there was no shift of the bands. 



20. In these early trials, however, I was not sure of seeing 

 more than one-tenth of a band shift, and with bisulphide of 

 carbon not more than one-fourth of a band. 



The bisulphide was not often employed, since water was 

 so much easier : water is not nearly so sensitive to changes of 

 temperature. Moreover the strong refractive index of bisul- 

 phide of carbon made it difficult to keep the beams to their 

 path, and prevent the bobbins from acting like prisms. But 

 the chief difficulty lay in the extreme sensitiveness to changes 

 of temperature, so that iioav the water-jacketing failed to 

 protect the bisulphide of carbon from some incipient elevation 

 of temperature, and the slightest change of that sort began 

 to contort the bands and render them undecided; while, if 

 the current was left on any considerable fraction of a minute, 

 they became unrecognizable. The experiment with the 

 bisulphide, therefore, could not be pushed to the same degree 

 of completeness as in the case of air or water ; but while the 

 observations lasted the effect was negative, and after each 

 experiment several hours had to be allowed for the liquid to 

 attain a uniform temperature again. 



There was no special theoretical reason for supposiug that 

 the magnetic drift would be any stronger in bisulphide than 

 in air, so this experiment w T as not persisted in. 



If a transparent magnetic liquid could have been used, per- 

 sistence w r ouid have been called for, but no such liquid is 

 known ; and though several attempts were made to use 

 a solution of some salt of iron, the minute increase of 



Phil. Mag. S. 6. Vol. 13. No. 76. April 1907. 2 M 



