698 
Proceedings of the Royal Society 
signs, gives the absolute expansion, in the direction of its length, of 
the glass rod, when the pressure at its surface is reduced by the 
observed amount, and consequently also of the compression when 
the process is reversed. As, in the case of non-crystalline bodies 
like glass, there is no reason why a given pressure should produce a 
greater effect in one direction than in another, we may, without 
sensible error, put the cubical compression at three times the linear 
contraction for the same pressure. 
The rod experimented on was made of lead glass, drawn by 
Messrs Ford of Edinburgh, and was 75*05 inches long. The tem- 
perature of the water in the hydraulic machine varied from 12*5° 
to 13*5° C. The pressure varied from 1 to 240 atmospheres. Ninety- 
one separate observations were made, and the general result is, that 
the linear compressibility of the glass under experiment is 0*96, and 
its cubic compressibility 2*92 per million per atmosphere. 
5. Suggestions on the Art of Signalling. By Alexander 
Macfarlane, M.A., D.Sc., F.RS.E. 
{Abstract.) 
After considering the analogy which exists between the arts of 
writing and of signalling, the author proceeded to discuss what 
alphabet is the most suitable where the physical agent is not 
electricity. If we choose for elementary signals two qualities of 
the agent of communication A and B, which can be produced in- 
dependently of one another, then the agent can be put into four 
states, viz., 1st, having the quality A, but not the quality B ; 2d, 
having the quality B, but not the quality A ; 3d, having both the 
qualities A and B ; 4th, having neither of the qualities A and B. 
One of these states is required to separate letter from letter, and word 
from word ; the fourth state where the agent is undifferentiated, is 
the one naturally adapted for the purpose. From the remaining 
three states we can get 3 permutations of one signal, 9 permutations 
of two signals, 27 permutations of three signals. Without going to a 
higher permutation than that of three signals, we get 39 symbols,* 
* Three of these would probably require to be omitted as repeating the same 
signal three times. 
