1888.] J. T. Bottomley on a Practical Air Thermometer. 93 
rausch, as to determination of the boiling point, the “ ice point ” 
merely being determined. An experimental determination of each 
point is, however, absolutely essential. 
Addition, Juneh, 1888. 
Shortly after the reading of the foregoing paper, I commenced to 
use the coal-gas oxygen blowpipe — employing Fletcher’s oxygen 
blowpipe and oxygen supplied in steel cylinders by the Scotch and 
Irish Oxygen Company (Brin Process). For convenience these 
cylinders, with the automatic apparatus supplied by the company 
for reducing the pressure of the gas, leave nothing to be desired ; 
and the use of the oxygen blowpipe makes easy and simple many 
operations which were formerly all but impossible. In particular, 
the working of Bohemian tubing becomes, without the slightest 
exaggeration, as easy as that of common flint or soft German glass ; 
and in addition it is a perfectly simple matter to make a junction 
between flint glass and Bohemian glass tubing (Bohemian glass does 
not join well with soft German tubing). Another great advantage 
in the use of oxygen with the Bohemian glass is, that the glass does 
not become porcelainised when worked with this flame, as it does 
when worked with the ordinary flame. 
With this new power to assist I have now abandoned completely 
the form of gauge shown in fig. 3, and instead 1 am using a gauge 
in which the main part is made of flint glass (stop-cocks of Bohe- 
mian glass cannot, so far as I know, be procured), but in which the 
air bulb a and capillary tube e are made of Bohemian glass, and the 
two glasses joined together a little below the bend at the top of the 
tube dd. I have not yet been able to obtain from any of the first- 
class makers of Bohemian tube a supply of fine capillary tubes, but 
this I make for myself by fusing up a piece of thick wide Bohemian 
tubing and drawing it down. 
2. On the Roots of gy Gustav Plarr, Docteur 
es-Sciences Math. Communicated by Prof. Tait. 
The imaginaries of Algebra have done good service during the 
process of discovery of the principles of quaternions. Now that 
those principles have been founded on the basis of operations on 
real lines, we must no more admit ^ - 1 as the equivalent of a 
