1183 
hand AR, 
No 
VE A Ae ce Dee Ags 
so that (47) is satisfied. 
When we take ue< 0, then we write for (46) in order to make 
the first term positive [viz. u, <0 and a; <0]: 
{ly dy By Ho + Gee wrd est Aa rl Beets (49) 
Consequently we equate now : 
{is < 
= he 
Us % 
Now we have: 
: ; up u “(up — Ug) : 
4p — 44 = ——— — > = SS : (50) 
Up—% Ug—% (up — *)({tg—*) 
As uw, is taken negative, # can be in accordance with (44) as well 
positive as negative; we now give to x one of the many negative 
values, which satisfy (44). With the aid of (44) and (50) we then 
find that again (47) is satisfied. 
Consequently we find: when the last group of a series is negative, 
then we may place this, after reversing its sign, before the first one 
and combine them to one single group; also it is apparent im what 
way we can find the new coefficients. 
We can still put the question whether all pairs of reaction- 
equations, which we can deduce from (1) and (2) will have the 
same series of signs. As a P,7-diagramtype is perfectly defined by 
its series of signs and reversally the series of signs is perfectly 
defined by a P,T-diagram, consequently this must be the case. When 
we deduce, therefore, from (1) and (2) another pair of reaction- 
equations, then the series of signs for this latter pair must, therefore, 
be the same as that for the first. Let the series of signs of the 
reactions (1) and (2) be given by (17), then this is also valid for 
each other pair of reaction-equations which can be deduced from 
(1) and (2). Of course it is possible that this new series of signs 
begins with another group; the order of succession, however, remains 
the same. In (17) the series begins with group A; when a new 
series begins e.g. with the group S, then is the order of succession; 
Se aCe | ats Ie DON eae Tc NE 
pasta Ba eRe ea Del: ree CA 
Oe 
In the first series the signs of the groups S, C, 7, and 7 are 
the reverse of those from (17), in the second series this is the case 
or 
