r364) 



Hk. J). 



l)iil lliuuuii 1 lia\e i-oiicliided lo llic cdurse ol' lliis \;iliu' of I) ..ƒ 

 above lueiitioiied from coiisideratioiis dorixed IVoiii ihis figure, 1 liave 

 inidei'stood afterwai-ds, lliat \ve may aeciiiire a siirvev of tliiscoui-se 

 in a simpler way. We may g•^^■e IT.,/ a somew lial diiiereiit sliape, 

 wliieli oeeiirs on \). i ('out. II, viz. : 



HV= /' + 



or 



yy :/; 



iV+ (Mi- 



Aboui the ([uantity (6s/)'j we kno\v, tiiat it is uegati\e, save for 

 the exceptional ease of water below 4°. ^) Of T,/ we know, that 

 inside the locus for which this quantity is equal to zero, it is positive 

 and increases rapidly, till it is infinitely large on the curve of the 

 ])oints D and J)'. And as the lactor of IV is necessarily positive, 

 it follows, that H',,/- is equal to zero on a curve, which lies between 

 That for which 1^" i^ e{pial to zero, in \vhich case ]]\,f = {s^.f)^ and 

 negative, and the curve, on which IT,/- is positive and has risen to 

 infinitely large. The latter curve is that of the [Hunts /) and JJ'. 

 We have come to the conclusion that the curve I^,/-=() in the 



1) See for the value of (e,/)y alóo "Ternary sy;>lemö III." These Proc. IV p. 08^. 



