440 KINSHIP AND ADAPTATION 
evolutionists believe that more or less modification is the 
rule among all the descendants of a given form, and that only 
rarely has it happened that ancestral forms have persisted 
for ages unchanged. Thus, in regard to the last examples, 
the most we should be willing to say is that mouse-tails were 
probably derived from plants something like the more primi- 
tive forms of recent buttercups, such as the ditch crowfoot 
(Fig. 209); that these ancestral crowfoots came from plants — 
something like our marsh-marigold; and so on. It is thought 
that single traits or a few traits in combination, are more 
apt to survive through long periods than the many peculiari- 
ties characterizing a complex structure. 
Closely connected with the erroneous view above men- 
tioned is the notion that all the modern representatives of a 
group can be arranged in a single series beginning with the 
most primitive, and passing on to the most highly evolved 
through survivals of the intermediate stages or ‘‘ connecting 
links.”’ From what has been said it will be obvious that such 
an arrangement would be possible only on condition that 
all the forms which had ever appeared were actually repre- 
sented by living descendants and that modification had 
occurred only in one direction. As a matter of fact, gaps, 
often great, between related forms are continually met with. 
Indeed, if the “connecting links” of the past, represented by 
the dead branches of our tree, had not disappeared there 
would be no possibility of classifying living forms into groups 
and subgroups, for there could be no limits to any group. 
Sometimes within a group it is as if Nature had as yet done 
little or no pruning, and the result is most bewildering to those 
who attempt a classification of the forms. No two students 
of the group are likely to agree as to where lines of demarca- 
tion should be drawn. Such a group is that of the roses 
previously mentioned. But even here, for all the embarrass- 
ing wealth of connecting links, it is quite impossible to ar- 
range the forms in a single series. There are many diverging 
series which branch again and again. Their relationships 
as inferred from their degrees of likeness can best be expressed 
by a branching system, and while of course the systematist 
must deal with his groups one after another insimple sequence, 
