18 DR BREWSTER on the Existence of Two New Fluids 



an elliptical bubble, and finally vanishes. When this takes place, 

 the boundary b b has of course disappeared, and d d and a a 

 have advanced to d'd' and a! a', and c c is invisible, in conse- 

 quence of the new fluid having spread over it, as it were, in the 

 manner described in the following section. 



Another cavity, consisting of three separate portions, AB, 

 CDE, FGHK, is shewn in Fig. 13., and is remarkable, in conse- 

 quence of each of these masses being connected with the ad- 

 jacent one, by a portion of the second fluid, which moves be- 

 tween them like a piston through the extremely narrow chan- 

 nels BC, EF. As the portion of new fluid between a b and ef 

 expands without having an air-bubble, it pushes the portion of 

 the second fluid B a b through BC into C a' b'. In like manner, 

 the second fluid c c? EF c' d varies its position with the expan- 

 sion of the fluids on each side of it. When the vacuity V dis- 

 appears, a portion of the second fluid shews itself in the space 

 D k h, and it again withdraws itself when the vacuity V touches 

 the sides of the triangular cavity. 



In some cavities where there is a large proportion of the se- 

 cond fluid, the vacuities sometimes form two-thirds and even 

 three-fourths of the space occupied by the expansible fluid when 

 the cavity is full, and yet these vacuities are filled at the usual 

 temperature of 83. In these cases, the circular vacuity did not 

 contract by heat, but extended itself till it disappeared. This 

 effect admitted of an easy solution, by supposing the surface of 

 the fluid to rise gradually by expansion ; but I found, by opti- 

 cal observations, that the vacuity occupied the whole thickness 

 of the cavity, and that it vanished by extension, when it was 

 held in a vertical direction. This remarkable fact will be fully 

 explained in the 5th section. 



In some specimens, the faces of the cavities are accidentally 

 inclined to the surfaces, nearly at the angles of total reflection 

 from the surface of the new fluid, so that all the part of the ca- 



