114 Proceedings of the Royal Society of Edinburgh. [Sess. 
called the Suspension Error ; it would arise, for example, from any shift in 
the point of suspension or from faulty correction for temperature; the 
terms (Aa/a){fA 1 a 2 + . . . } may be called the Circular Error, reserving, how- 
ever, the question whether they are adequately represented by taking 
X 1 = — <y, Ag = 0 as is done in the ordinary theory ; the other terms 
K 
— • a 
n 
riD 
cla 
Aa 
a 
- I3A 
n 
are naturally called the Escapement Error. We see, then, that there are 
five variables, An, Ad, Aa, AW, A(k/u), and that variation of rate cannot be 
defined in terms of less than three of them, treated as independent. To 
examine how far the actual going of the clocks can be shown to agree or 
disagree with this theory would require an analysis of observed rates, as 
well as specially directed experiments that do not form a natural part of 
the present paper. It is reserved for a future study. I would mention in 
conclusion that the Barometric Error has, for simplicity’s sake, not been 
expressly referred to, but it will appear in due course as a natural extension 
of the theory given. 
( Issued separately June 17, 1918.) 
