IN PLATES, TUBES, AND CYLINDERS OF GLASS. 363 



With the view of confirming these results, I took a file with 

 a very sharp edge, and cut the tube entirely through by a notch 

 EF, Fig. 5. By this operation the particles were freed from 

 the state of violence in which they were held, and assumed 

 the very same arrangement which they never fail to take in 

 rectangular plates of glass. By exposing the tube, thus divi- 

 ded, to polarised light, it exhibited the appearance shewn in 

 Fig. 5. where mp n o, rri ' p ' n' V, are two dark fringes having a 

 negative structure on the outer side of each, and a positive 

 structure between them, as in plates of glass with two axes. I 

 obtained the same result with a tube of a very large bore, ha- 

 ving its exterior diameter 0.875 of an inch, and its interior dia- 

 meter 0.816 ; so that the thickness of the glass was only 0.059 

 of an inch. When the tube is cut into two or more pieces, each 

 piece has the same structure as a plate of glass with two axes. 

 By a number of measurements, I have found that the diameter 

 op, Fig. 3. of the black fringe or circle of no-polarisation, is 

 a geometrical mean between the interior and exterior diame- 



ters of the tube, that is, op := a/ AB X ab . 



If a tube of glass is brought to a red heat, and then cool- 

 ed by inserting in its bore a cylinder of iron, or any other 

 conducting body, the structure will then be the same as is re- 

 presented in Fig. 5. 



If a solid cylinder of glass which has only one structure, is 

 perforated in its centre, it will exhibit the appearance in Fig. 3,, 

 and if it is divided by a notch, it will acquire the structure 

 shewn in Fig. 5. 



When polarised light is transmitted through a long cylinder 

 of unannealed glass, in a line perpendicular to its axis, after 

 it is immersed in a fluid of the same refractive power, it exhi- 

 bits the same phenomena as rectangular plates of glass, having 

 a positive axis perpendicular to its length, and a negative axis 

 perpendicular to the positive one ; but in the cylinder, the ex- 



Z z 2 ternal 



