Hull, Rainbows in Retrospect 
531 
phylogenetics. Without realizing it, Hennig had set out the general principles of 
natural classification. Cladistic methods discover timeless, static patterns, and patterns 
are necessarily prior to the processes that produce them. 'Pattern analysis is, in its 
own right, both primary and independent of theories of process, and is a necessary 
prerequisite to any analysis of process' (Nelson and Platnick 1981: 35). 
Although Hennig did not distinguish clearly and consistently between cladograms 
and trees. Nelson and Platnick (1981:141-42) do. Cladograms have no necessary 
connection to evolution. Cladograms represent patterns within patterns, and this 
concept of patterns within patterns is an 'empirical generalization largely independent 
of evolutionary theory, but, of course, compatible with, and interpretable with 
reference to, evolutionary theory. The concept rests on the same empirical basis as 
all other taxonomic systems (the observed similarities and differences of organisms).' 
Nelson and Platnick (1981:142) go on to note that two of the basic elements of 
cladistic analysis (relationship and monophyly) are 'definable only with reference to 
the branching diagram, and carry no necessary evolutionary connotation.' However, 
the concept of patterns within patterns is 'not wholly independent of evolutionary 
theory' because its third element (synapomorphy) is commonly interpreted only in 
connection with evolution. But, if 'synapomorphy' is defined 'purely as an element 
of pattern — a unit of resolution, so to speak,' then cladistics becomes the 'general 
theory of taxonomy of whatever sort' (see also Nelson & Platnick 1981:165). Parallel 
observations hold for the term 'homology' as well (Nelson & Platnick 1981:159). 
Thus, 'cladistics' in the broad sense concerns the general methods of discovering 
patterns within patterns, while 'cladistics' in the narrow sense is limited to the study 
of those patterns that arise through the evolutionary process. With respect to cladistics 
in the broad sense. Nelson and Platnick (1981:324) note that some persons 'may 
think it strange to use words beginning with "dado-" in a sense divorced from 
evolution and phylogeny' but go on to argue that such usage is justified given the 
etymology of the term. 
Reference to 'cladistics' in a broad and narrow sense may sound reminiscent of 
Sneath's (1995:281) distinction between 'numerical taxonomist' in the broad and 
narrow sense, as it should. 'Numerical taxonomy' refers both to the use of quantitative 
techniques in systematics and to numerical phenetics. 'Cladistics' refers both to the 
general principles of the recognition of patterns within patterns and more narrowly 
to the applications of these principles to phylogeny. 
As in the case of pheneticists, cladists reject the view that they ever held that 
classification can be or should be theory free or theory neutral. From the start. 
Nelson and Platnick (1981: 301) acknowledged that a character is a 'theory that two 
attributes which appear different in some way are nonetheless the same (homologous). 
As such, a character is not empirically observable, and the hope of pheneticists to 
reduce taxonomy to mere empirical observations seems futile' (see also Platnick 
1979: 542, 1985:88). 
By now, it should be clear that the notion of theory-free classification is far from 
clear. Both the pheneticists and the pattern cladists claim never to have held the anti- 
theoretic views attributed to them. On this score, Johnson and I are both equally 
guilty because both of us thought that pheneticists and pattern cladists had something 
against letting 'theories' enter into the classificatory process, especially in the early 
stages. As it turns out, the issue is merely which theories are to enter into the 
classificatory process right from the start. As Platnick (1985: 88) sees it, 'phenetics is 
no more theory-free than is cladistics — it's just based on a different theory' (for 
more recent views on pattern cladism, see Grande & Rieppel 1994). 
