80 
THE AVONIAN OF THE AVON GORGE 
to north, the entrances of the coral gentes occur later and later, 
as measured over a considerable segment of the Brachiopod- 
scale, whereas the sequence of variation-incidents, which form 
the graduations, remains constant for both the Brachiopod and 
Coral scales.) 
This irrationality of two time-scales, constructed from the 
variations of two different classes, represents a relative accelera- 
tion of the one class upon the other and its existence puts a 
definite limit to the smallness of useful zonal divisions ; for the 
size of a zone or subzone must necessarily be large when com- 
pared with the error which may be introduced by relative 
acceleration. 
We have now to tabulate the successive species or mutations 
of each gens of the two classes we have selected and to express 
the relative times at which each mutation was dominant. 
It has been pointed out by several observers that a species, 
during its period of dominance, retains its distinctive characters 
without appreciable change and that transitional forms con- 
necting it to other species are relatively scarce. 
This general principle is strikingly exemplified in the variation 
of a gens. Each new mutation arrives almost unannounced 
and is usually never more abundant than at the time of its first 
establishment. (Careful search usually results in the detection 
of a few early forerunners of each mutation and, in a small 
number of cases, the maximum development of a species does 
not take place until some time after its first establishment.) 
Each gens may be conveniently represented graphically by a 
line upon which stations are marked which represent the maxima 
of successive mutations, and the distance between two stations 
is measured proportional to the actual thickness of strata in the 
Avon Section which intervenes between the maxima of the two 
mutations. (Where a station cannot be fixed from the occur- 
rence of a mutation in the Avon Section itself, its position must 
be approximately fixed by correlation.) This method is of more 
practical use than the attempt to plot accurate time-intervals 
