in the Velocity of Propagation of Light in Water. 277 



r was filled with very fine cotton- wool. The shutter K always 

 remained closed until the commencement of the experiment, in 

 order to diminish as much as possible the radiation of heat and 

 the cooling of the glass plates. 



When I began my measurements with the apparatus con- 

 structed in this manner, the inconvenience at once declared itself 

 that, in order to maintain the flat glass plate tight up against 

 the glass tube, it was necessary to screw the three screws very 

 hard. In consequence of the great pressure to which the plates 

 were thereby subjected, the images of Fraunhofer's lines at 

 higher temperatures appeared indistinct, and the reflexion of 

 distant objects showed either no images at all or only extremely 

 indistinct and bleared ones, so that the desirable accuracy could 

 not be attained in the angular measurements. When, however, 

 finally, and with the greatest labour, a series of measurements 

 had been taken, they showed such irregularities as to make it 

 certain that a considerable source of error must be present. 

 When, in consequence of this, the alteration in the prism by 

 heat came to be examined by the scale and telescope, these alte- 

 rations were not only found to be more considerable than was 

 expected, but also to vary in no manner connected with the 

 temperature. 



The entire apparatus was taken to pieces ; and the cause of 

 these remarkable alterations in the angle was found to be owing 

 to a mistake of the mechanician, who had allowed one of the 

 glass plates to touch the brass of the box P t . The defective re- 

 flexion arose from the glass plate, by projecting a little over the 

 glass tube, and being fastened, as is shown, fig. 6, assuming a 

 convex form when the pressure was increased. An entire re- 

 arrangement of the apparatus became necessary; and the obser- 

 vations obtained, the result of several weeks' labour, had to be 

 rejected. In place of the glass tube a brass one was employed, 

 which was soldered into P, a brass collar being soldered on to Pj 

 at the place where the tube entered it. This brass was ground 

 flat with extreme care, so that the flat glass plates only required 

 to be pressed very gently in order to secure a perfectly water- 

 tight joint. Upon the glass plates was laid a caoutchouc ring, 

 and upon this a brass diaphragm, which was pressed very gently 

 by means of the screws (fig. 7). In this manner not only was 

 the joint tight, but every alteration in the form of the glass was 

 avoided. In order to prevent the oxidation of the brass by the 

 liquids introduced or by the air, the whole of the inside of the 

 tube was strongly gilded. All the experiments were performed 

 with this arrangement of the instrument, which proved itself a 

 thoroughly good one. 



It is true that slight alterations are shown in the angle with 

 Phil. Mag. S. 4 Vol. 35. No. 237. April 1868. U 



