Hull, Rainbows in Retrospect 
535 
I am afraid that I am at a loss to figure out what Sneath and Sokal intend. If a 
systematist has made generalizations about taxa at the end of his first run through of 
the steps that Sneath and Sokal list, is he allowed to use these generalizations when he 
returns to step one? If so, then this method looks like the much condemned vertical 
construction of hypothesis upon hypothesis. If not, then all the systematist is doing is 
junking the results of his first run through and starting all over again from scratch. 
Pattern cladists took as part of their inspiration the writings of Karl Popper. 
(I suspect that Syste7natic Zoology is the only journal that includes reviews of all of 
Popper's major works.) One of Popper's fundamental positions is that scientists 
need not and cannot begin their investigations solely on the basis of theory-free 
observation statements. Even the most observational of terms include all sorts of 
theoretical assumptions built into them. 
For example, Johnson (1976:160) lists the sorts of morphological features that he 
used to discriminate species in the genus Eucalyptus. They include: 
... more-or-less gross characters of the bark, leaf-shape, opposite and decussate 
versus 'alternate' (actually pseudoalternate and still essentially decussate) 
phyllotaxy, shapes and sizes of peduncles and of flower parts (especially the 
operculum and the anthers) and details of size and shape of the fruits. 
Anyone familiar with the history of botany is aware that what counts as bark, 
leaves, peduncles, opercula, anthers and fruits, not to mention various sorts of 
phyllotaxy, are not matters of simple observation. No ordinary person could simply 
look at a tomato and see that it is a fruit. Very complicated arguments and theoretically 
committed distinctions go into defining the notion of what is or is not a 'fruit.' 
Defenders of theory-free classification respond at this point that the characters listed by 
practicing systematists way well be highly sophisticated, theory-impregnated 
constructions but that these concepts can be analyzed into absolute simples which are, 
in the relevant sense, theory free. Ultimately the entire observational basis of science can 
be replaced by statements such as 'Red spot now.' Of course, terms like 'red' and 'now' 
depend on physics, but for biological classification such dependence is unproblematic. 
As seductive as this position may seem, it has proven to be a total failure. In Der logische 
Aufbau der Welt, Rudolf Carnap (1928) proposed to reconstruct the entire world of our 
experience in terms of absolute observational simples. However, he was unable to 
produce a single theory-free observation statement that satisfied even himself. 
Pattern cladists now claim that they never thought that theory-free observation terms 
exist, numerous misleading statements not withstanding. Quite obviously, the view 
that pheneticists and pattern cladists have been trying to enunciate is extremely 
subtle, possibly too subtle to be expressed in any natural language. Or just possibly, 
they have changed their minds on this topic. Because I was taught in my early days 
in biology that one of the strengths of scientific investigation is that it can force you 
to change your mind, I continue to be puzzled by how resistant scientists (not to 
mention philosophers and theologians) are to admitting that an earlier view that 
they held may well be mistaken, especially if it is fundamental to their entire world 
view. 'Oops' is not a popular word in scientific publications. 
Sokal and Sneath are willing to re-evaluate and even abandon some of their early 
views (e.g., matches asymptote and non-specificity), but overall-similarity and general 
purpose classifications are quite another matter. Either they must remain untouched 
or else be transformed surreptitiously. Cladists are willing to abandon several of 
Hennig's principles but his basic methodology of three-taxa statements remains 
inviolate. The 'theories' that underlie cladistics are not theories about what is or is 
not a fruit, but what is or is not a character. 
