PHILOSOPHICAL FOUNDATIONS OF CLASSICAL EVOLUTIONARY CLASSIFICATION

WALTER J. BOCK

Abstract Bock, W. J. (Department of Biological Sciences, Columbia University, New York, New York, 10027) 1974. Philosophical foundations of classical evolutionary classification. Syst. Zool. 22:375-392.—The primary objective of biological classification is to provide the

foundation of all comparative studies in biology. These studies require consideration of Downloaded from two factors, namely: (a) the degree of genetic similarity between organisms; and (b) the phylogenetic sequence of events in their history. Classical evolutionary classification provides the best approach to classification, based on Popper's criterion of content, because it attempts to maximize simultaneously both of these semi-independent variables. Evo- lutionary classification is based on the evolution of organisms, not just their phylogeny. Application of the Popperian philosophy of the demarcation and methodology of science to classification suggests that a major task is the development of severe tests of falsification

by which classifications can be tried. Although phenetic methods appear to be the best http://sysbio.oxfordjournals.org means of recognizing taxa and deducing classifications, cladistic methods appear to pro- vide the best tests in attempts to disprove these statements of relationships. The best approach to definitions in this historical science is a theoretical one in which the defining criteria and the recognizing criteria may differ. Comparisons, which are closely akin to definitions, must be based on evolutionary theory. Homology is the primary principle in comparative biology with the possibility that all other principles of comparison are re- ducible to homology. The defining criterion of homology is phylogeny and the only recognizing criterion of homology is similarities of all sorts between features. Any approach to classification that excludes homology and its recognizing criterion of similarity as a primary step in deducing relationships is invalid. Use of falsifying tests suggests that relative weighting of characters may have greatly reduced importance in taxonomic methods. Lastly, the formal classification and the phylogenetic diagram of a group of organisms are not different, but redundant images of each other. Both are essential parts of the conclusions of any study of biological relationships. [Philosophy; Evolutionary at Stanford Medical Center on June 7, 2010 classifications; Definitions; Comparisons; Homology.]

INTRODUCTION accepted classifications that accomodated Classification of living organisms dates readily the numerous new species of living back into antiquity, having its roots in the and fossil organisms being described daily, development of language and the first at- The maior deficiency in this early nineteenth tempts to convey information from one century systematics was a satisfying philo- person to another. The earliest organized sophical and scientific foundation. Biology efforts to classify animals and plants pre- was bf nS *?<& alon§ f™1/ in ^ *an- sition from e an ient idea dates the beginnings of oral and written * £ taxonomy (Sokal and Sneath, became the philosophical basis for classifi- 1963), are largely concerned with the cation. The notion that a group of organ- development of techniques by which the isms are related because of their evolution degree of similarity (or difference) be- from a common ancestor became the funda- tween organisms can best be measured. mental criterion underlying biological com- Arguments as to whether phenetists parison and classification. All major modern measure genetical or phenotypical similarity approaches to biological classification are or whether they compare "overall" or "over- Downloaded from evolutionary because they are based on the some" similarity are of little significance. Darwinian theory of evolution and accept Degree of similarity between organisms the criterion of descent from a common is an important factor in ascertaining ancestor as the basis for relationship. relationships and in establishing classifica- Simple as this latter statement may be, it tions; phenetists are correct in stressing its is subject to several interpretations which importance. Most significant is that in pure http://sysbio.oxfordjournals.org have given rise to widely differing ap- phenetics, the only variable used in con- proaches to classification. Analysis of these structing classifications is similarity between varying interpretations of common descent organisms. No phyletic information is used as the hallmark of relationship has been a (Colless, 1967), this being considered by central theme of classificatory theory. phenetists as "the phylogenetic fallacy." The approaches to biological classifi- Although I believe that this fallacy is in- cation arising from the varied interpreta- valid (Bock, 1969 b), I agree with phenetists tions of common descent can be grouped that phyletic information must be excluded under three headings which characterize from their approach to classification. As the major modern schools of systematics. soon as any phyletic information is ad- The purpose of this essay is to appraise the mitted into phenetics, then no basis exists at Stanford Medical Center on June 7, 2010 philosophical foundation of only one of to exclude any phyletic information, with these approaches—that of classical evo- the result that the philosophical foundation lutionary classification—not to contrast of phenetics has been modified radically, critically or evaluate all three. In view of and little would separate phenetics from their common origins in Darwinian evo- classical evolutionary classification. Such lution, brief demarcation of each school is problems exist with the development of essential to insure clarity of discussion and techniques such as phenetic cladistics meaning. These schools of systematic which have emerged from pure phenetics theory are: without a careful restatement of its philo- A) Phenetics: This approach is often, sophical base. but erroneously, synonymized with numeri- B) Cladistics: The term "cladistics," cal taxonomy which is a methodology often referring to branching, is less ambiguous used in phenetics but has broader useful- and indeed more descriptive than the term ness. Phenetics is based on the concept that "phylogenetic systematics" (Hennig, 1966) relationship between organisms is ascer- which is preferred by most proponents of tained by their overall degree of similarity. this approach to classification. Relationship Greater similarity indicates closer relation- among organisms, in cladistics, is based ship and a smaller amount of evolutionary upon joint possession of derived features change from the common ancestor. Implicit ( = apomorphous or advanced characters). in this approach is the assumption that the Species which share derived features are degree of difference between organisms is related, and hierarchies of taxonomic proportional to the amount of evolutionary groups are established on hierarchies of change since their common ancestor and derived characters. Thus relationships be- hence to their degree of relationship. tween organisms are strictly determined by I.C.S.E.B.—SYSTEMATIC PHILOSOPHIES 377

phylogenetic branching and may be re- The name for this third approach to garded as genealogical or kinship relation- biological classification is awkward and ships. somewhat misleading in that it carries the Cladistics is based strictly upon joint implication that this is the only approach possession of derived features and excludes based on organic evolution. This is not the absolutely any evaluation of similarity from case, and once again it should be empha- the determination of relationships and con- sized that all major approaches to classifi- struction of classification. Thus, Hennig cation are based on Darwinian evolution. states clearly (1966:10) that "Genealogic Evolutionary classification is an eclectic Downloaded from relationships are, however, something en- approach in that it combines the important tirely different from 'similarity,'" and (p. elements from phenetics and cladistics, but 12) "The second error in the idea of a "eclecticism" is a poor name because it is logical and historical primacy of nonphylo- not descriptive of the basic conceptualiza- genetic (e.g., typological) systematics stems tion of this approach to classification.

from the assumption that in biological Evolutionary classification is based upon http://sysbio.oxfordjournals.org systematics the primary relationships be- a simultaneous evaluation of the two vari- tween living entities are similarity re- ables used singly by phenetists and cladists lationships (especially those of structural to ascertain relationships between organ- morphology), and that the elements of isms, namely: systematic work should be individuals, if 1) Evaluation of the amount (degree) of not even species." Rejection of similarity genetical similarity between organisms as as a factor in classification is repeated in judged by the degree of their phenotypical his third and fourth points (Hennig, 1966: similarity. Greater phenotypical similarity 23) summarizing the reasons for favoring implies greater genetical similarity and cladistics as a general reference system for hence closer relationship. Greater similarity

biological sytsematics. between species is indicative of a smaller at Stanford Medical Center on June 7, 2010 The importance of phyletic information amount of evolutionary modification of in classification is correctly emphasized by each species from their common ancestor cladists. Not only is this an important and hence possession of a greater shared variable in any general reference system in amount of genetical material. biology, but the best ( = most severe) tests 2) Evaluation of the sequence of events for disproving taxonomic statements are in the evolution of each species from the based upon phyletic information. And, as common ancestor, with the events arranged stated above for phenetics, pure cladistics in phylogenetic order. This includes proper is based upon consideration of only phylo- sequential arrangement of the origin and genetic relationships judged by common modifications of features and of the branch- possession of derived features. Inclusion of ing sequences of each lineage. Common similarity in a cladistic approach to classifi- possession of derived features in different cation would erase the difference from evo- organisms implies common ancestry with lutionary classification and cannot be done, hierarchies of phylogenetically arranged as recognized by its proponents, without derived features used to establish taxonomic abandoning the philosophical foundation hierarchies of groups. of cladistics. (Rejection of similarity as A general correlation exists between the an important factor in the determination degree of similarity between organisms and of relationships and establishment of classi- the phylogenetic recency of their common fication results in a serious internal flaw in ancestor, but this is not an exact correlation. cladistic philosophy and methodology as Otherwise we would not have the contro- shown below in the discussion of versy that exists today between different homology.) schools of taxonomic thought. The degree C) Classical evolutionary classification: of similarity and the phylogenetic sequence 378 SYSTEMATIC ZOOLOGY of events must be regarded as two semi- methodologies. A careful reading of the independent variables (i.e., only partly cor- literature of classical evolutionary classifi- related with one another) with the degree cation and consideration of the general of independence varying from group to acceptance of the classification of animals group and from time to time within any and plants—how well they endured and particular group. In evolutionary classifi- were able to accommodate new biological cation, relationship between organisms, as discoveries—testifies to the existence of a

expressed in formal taxa, attempts to maxi- viable philosophy and a sound method- Downloaded from mize simultaneously both semi-independent ology. Most evolutionary taxonomists knew variables of degree of similarity and phylo- what they were doing and why, even in the genetic sequence of events. No reason absence of a clear statement of philosophy exists why biological classification must be and methods. This lack of attention to the based on only one variable and why it philosophical basis of a science is not cannot be based on two (or more) variables, unique to biological classification and is be they semi-independent or completely not necessarily a serious determent to rapid http://sysbio.oxfordjournals.org independent. A major problem is that the progress as shown by Reichenbach (1968: independence of these variables varies in 114) for physics during the nineteenth peculiar ways; hence, taxonomists are century. forced to make difficult decisions based A firm philosophical foundation for upon evaluation of many factors. The same classical evolutionary classification has ex- decisions will not be reached by all taxono- isted for many decades, but has not been mists and hence the charge of subjectivity expounded as clearly as it deserves. The leveled at evolutionary classification by purpose of this essay is to outline briefly proponents of other approaches. Such a (a) the basis for judging which approach charge may be true, and I will not argue to biological classification provides the best it, but simply want to emphasize that this general reference system and (b) the at Stanford Medical Center on June 7, 2010 subjectivity is a consequence of basing philosophical foundation for classical evo- classification on the simultaneous evalu- lutionary classification. ation of two semi-independent variables At the onset, I would like to acknowledge rather than just one of the two. The result- my debt to Ernst Mayr and George G. ing classification based on this eclectic Simpson who have done more to clarify approach may be "subjective" in the sense the philosophical foundations and method- that not all taxonomists will automatically ologies of classical evolutionary classifica- reach the same conclusion, but it has the tion than any other living taxonomist. advantage, as will be argued below, of pro- Almost all of the ideas expressed in this viding a more realistic picture of the com- essay can be found readily in the writings plexities of the biological world. of these taxonomists, such as in their I prefer the descriptive adjective "classi- "Principles of animal taxonomy" (Simpson, cal" before evolutionary classification be- 1961) and "Principles of systematic zool- cause I believe that this approach to ogy" (Mayr, 1969). Hopefully this paper biological classification had its roots in will serve as a guide to the more extensive systematic studies long before Darwin and discussions of the philosophy and principles continued after 1859 with a change in its of classification by Mayr and Simpson. philosophical foundation and basic theory as the major school of classification. During BASIS OF BIOLOGICAL CLASSIFICATION its development, more attention was given Foremost, in discussions of the varied ap- to practical considerations—the description proaches to biological classification, is the and classification of organic diversity— question—which of the possible schemes than to the clarification of philosophical provides the best classification? This may foundations and the precise recording of be expressed in terms of "a general refer- I.C.S.E.B.—SYSTEMATIC PHILOSOPHIES 379

ence system" (Hennig, 1966:9) which ex- history means precisely the exact pattern presses well one of the essential purposes and sequence of branching of phyletic of biological classification. Needless to say, lineages. This criterion, which is the goal proponents of each school of systematics of cladistic approaches and, in part, of believe that their approach provides the evolutionary approaches to classification, is best classification and provide arguments a significant aspect of classification but one supporting their position. Unfortunately, of considerably less important direct use to

claims about the best classification or gen- biologists when employing classification in Downloaded from eral reference system are vague unless firm their research. The phylogenetic criterion criteria are established first on which to is of prime concern in the analysis of the judge different classifications, as has been evolutionary history, including mechanisms, emphasized by Warburton (1967). A set of individual features and organ systems of criteria must be stated clearly, even if and in many biogeographic studies. not universally accepted, before different Other criteria have been suggested such classifications can be judged best relative as that classification should be the most http://sysbio.oxfordjournals.org to one another. accurate representation of the evolution of The most fundamental attribute of bio- organisms, or that classification should be logical classification is that it must be a natural, realistic or holistic system. These useful. It must order and summarize bio- other criteria either can be reduced (e.g., logical information, must be heuristic, and the evolutionary criteria) to either one or a must provide the basis for future studies. If combination of the two criteria mentioned a classification does not meet these mini- above, or are (e.g., the natural system) so mal utilitarian requirements, it should be vague that judgement of the best classi- rejected forthwith. fication is impossible. Several sets of criteria have been sug- The several criteria, namely maximum gested as a basis on which to judge the best predictability and phylogenetic history, at Stanford Medical Center on June 7, 2010 system of classification. One widely ac- commonly suggested for evaluating the best cepted criterion is that of greatest predict- classification are important attributes of ability of unknown characters in known any system of relating organisms. Yet, they organisms or in newly discovered species are subordinate components of a more (Warburton, 1967). Thus the best classi- fundamental basis of biological classifica- fication is based on the greatest concur- tion which should serve as the criterion for rence of characters in organisms which evaluating the best system of classifying should permit the most accurate prediction organisms. of unknown characters. This criterion The primary objective of biological embraces an extremely important attribute classification is to provide the foundation of classification, one that is of prime con- for all comparative studies in biology—with cern to a large majority of biologists when the best classification being the one that using classification as a basis of their re- permits the most useful comparative in- search. Maximum prediction of unknown vestigations. Although this statement ap- features represents the general goal of pears as vague as claims that classifications phenetic approaches and of many evolu- should be natural, criteria can be estab- tionary taxonomic approaches to classifica- lished for judging the most useful compara- tion. tive studies in biology. A second criterion on which to judge the If one accepts the primary objective of best classification is the most accurate biological classification, to be the founda- representation of the phylogenetic history tion for all comparative studies, then one of organisms. This is the basis of consider- must accept the notion that only one sys- ing classification as the general reference tem of classification is possible and that system in biology by cladists. Phylogenetic this system must be the most optimal one 380 SYSTEMATIC ZOOLOGY for all comparisons. If biologists wish to which is logically stronger; which has achieve a unified science with a common the greater explanatory and predictive set of explanations for all biological phe- power; and which can therefore be more nomena, then the results of all comparative severely tested by comparing predicted biological studies must conform with each facts with observations. In short, we pre- other—they must be interconnected or com- fer an interesting, daring, and highly parable—otherwise each biologist would informative theory to a trivial one.

not be able to comprehend and use the re- All these properties which, it thus Downloaded from sults obtained by other biologists. To be appears we desire in a theory can be sure, some particular system of classifica- shown to amount to one and the same tion may serve as the best foundation for thing: to a higher degree of empirical a particular or specialized comparative content or of testability." study, but the results will be of less general The scientific theory that best meets these usefulness. Thus I reject the notion that criteria as the foundation for all biological different approaches to classification are comparison is the Darwinian theory of http://sysbio.oxfordjournals.org equally useful and that many different organic evolution. schemes of classification for a group of All attributes of living organisms, their organisms are possible and a worker can form and functions, their similarities and choose the one best suited for a particular differences, and their correlations and pat- study. Such a notion would simply increase tern of distribution, depend on the past the chaos in biology instead of ascertain- evolutionary history of these organisms, not ing the existing order in biological diver- simply on their phylogeny (see Bock and sity for which systematists have assumed von Wahlert, 1963, for a distinction be- primary responsibility. tween evolution and phylogeny). Hence, Comparative studies of biological organ- comparative study of biological attributes isms can be based upon a number of must be based on a full consideration of at Stanford Medical Center on June 7, 2010 different philosophical foundations. Non- their evolution, including all evolutionary scientific philosophies (e.g., ideal typology) mechanisms and phenomena that have can be rejected using the criterion for contributed to the development of the demarcation of science advocated by present condition of these features. If Popper. Yet it may not be possible to judge biological generalizations are attempts to which of the scientifically valid foundations summarize the results of organic evolution, of comparison are correct or wrong, and it then such an eclectic approach is necessary is probably unreasonable to attempt to do because these generalizations are depen- so. However, it is possible to distinguish dent upon the system of comparison used to between these different philosophical bases formulate them. for comparison and to choose the best one on using another set of criteria which have Comparative biological studies include been summarized by Popper (1968b: 217) several general types of questions which are interwoven extensively with each other. as: One common question is the similarity of "This criterion of relative potential satis- attributes of members of a taxonomic group factoriness (which I formulated some which bears on the degree of common time ago, and which, incidentally, allows genetical material, and hence the number us to grade theories according to their of phenotypical features possessed in com- degree of relative potential satisfactori- mon. This is associated with the correlation ness) is extremely simple and intuitive. of characters and the question of whether It characterizes as preferable the theory all members of a taxon will possess a new which tells us more; that is to say, the attribute discovered in one of them. A theory which contains the greater amount biologist is often concerned with the of empirical information or content; question of how many species, and which I.C.S.E.B.—SYSTEMATIC PHILOSOPHIES 381 ones, should be examined before having phenetics and cladistics are wrong. Rather, confidence in the generality of a newly I claim that using the criterion of a higher discovered attribute. An experimental biol- degree of empirical content or of testability, ogist or medical researcher must know evolutionary classification provides the best whether he can extrapolate the results of foundation for all biological comparisons. a study done on one species to another on Second, that this criterion for the best which he cannot experiment, and what are classification is the most essential distinc- the chances of being correct. Many phy- tion between classical evolutionary classi- letic questions are important, such as fication on the one hand and phenetics and Downloaded from whether many of the similarities between cladistics on the other. After completing mammals and birds exist because of their this manuscript and reviewing it in prep- common ancestry in reptiles or are the re- aration for the symposium discussion, I sult of both groups being warm-blooded; realized that most, if not all, of the remain- or whether attributes found in living am- ing comments on the philosophical founda- phibians can be assumed to be similar to tions of evolutionary classification apply http://sysbio.oxfordjournals.org those present in the earliest tetrapods? De- equally well to cladistics and phenetics. cisions on many such questions may involve That is, the basic ideas of cladistics and large cost factors in terms of money and phenetics could be reduced to the concepts time, and the results may be of utmost im- to be presented below although they may portance in terms of medical practice or not be so expressed by the proponents of agriculture. Thus the decision on the philo- either school of classification. Nor will sophical foundation of biological comparison all advocates of cladistics or phenetics and classification is not merely of theoreti- agree with the positions I adopt, such as cal interest to a small handful of sys- how definitions should be presented or the tematists, but is of extreme practical central role of homology. importance to everyone, biologist and at Stanford Medical Center on June 7, 2010 nonbiologist alike. PHILOSOPHY OF SCIENCE Classification, as the foundation for com- Biological classification is a science, and parative studies in biology, must include as such should be treated within some two factors, namely: (a) degree of genetical philosophical framework of the criteria of similarity as ascertained by the degree of scientific theories and methodologies. similarity of phenotypical features shared Herein, I accept the philosophical approach in common; and (b) the phylogenetic to science advocated by Karl Popper in sequence of events, including branching his "The logic of scientific discovery" of lineages, in the history of the organisms (1968a) and later in his "Conjectures and as ascertained by the shared possession of refutations" (1968b) and "Objective knowl- derived features among other methods. The edge" (1972). And I would advocate that best classification is the one that maximizes the theory and particularly the methodol- simultaneously both of these semi-inde- ogy of biological classification should be pendent variables, not just one of them. formulated according to the Popperian Classical evolutionary classification is the philosophy of science. approach, in my opinion, that achieves this Popper's basic philosophy of empirical simultaneous maximization of both vari- science may be summarized in two points. ables, and hence provides the best system First, that inductive methods in science of classification for all comparative studies are not valid. Rather that science may in biology. "be described as the theory of the deduc- Two points must be emphasized. First, tive method of testing, or as the view that that I do not claim that classical evolu- a hypothesis can only be empirically tested tionary classification is the correct approach —and only after it has been advanced." to biological classification and that (Popper, 1968a: 30). Second, that the 382 SYSTEMATIC ZOOLOGY demarcation of empirical science from "verification." A theory that has withstood other areas of human thought which re- many attempts to disprove it with severe quires criteria that separate science from tests possesses a high degree of corrob- metaphysics but do not exclude from the oration. These are not highly probable domain of empirical science statements that theories as Popper emphasizes (1968b: 58), cannot be verified. Thus Popper accepts namely that: "Although we seek theories as the criterion for demarcation of science with a high degree of corroboration, as the concept that scientific theories are scientists we do not seek highly probable capable of being disproven, not proven or theories but explanations: that is to say, Downloaded from confirmed. He states (Popper, 1968a:40- powerful and improbable theories." 41): Thus, scientific theories cannot be proven "But I shall certainly admit a system as or verified, but can only acquire a high empirical or scientific only if it is capable degree of corroboration after repeated of being tested by experience. These failures to falsify them. This is clearly seen considerations suggest that not the in any statement about relationships. http://sysbio.oxfordjournals.org verifiability but the falsifiability of a Biologists may have great confidence in system is to be taken as a criterion of some highly corroborated classification— demarcation. In other words: I shall not until someone disproves it in a severe test require of a scientific system that it shall using newly discovered evidence. be capable of being singled out, once and Popper's philosophical approach to for all, in a positive sense; but I shall science is not new; he served the important require that its logical form shall be such role of stating these ideas in a clear, that it can be singled out, by means of concise fashion and generalizing them into empirical tests, in a negative sense: it a unified philosophy of science. Ghiselin's must be possible for an empirical scien- discussion (1969a) of the hypothetico-

tific system to be refuted by experience." deductive method as used by Darwin is a at Stanford Medical Center on June 7, 2010 somewhat specialized analysis of the Thus, a classification, a phylogeny, a con- general philosophy stated by Popper. The clusion about homologous features or any "method of reciprocal illumination" advo- other statement about relationships be- cated by Hennig (1966:21) is a similar but tween organisms are scientific theories more specialized expression of the same susceptible to testing in the attempt to dis- concepts. Unfortunately, the Popperian prove them by empirical observations. philosophy of science has not been broadly Numerous statements in the systematic applied to classification even tacitly. In- literature that the available evidence proves deed, it is surprising how few systematists some particular relationship or, in a more are aware of these ideas; Mayr, and sophisticated sense, that affinities between Ghiselin (1969a, b) are among the rare organisms can be demonstrated with vary- exceptions. It is my belief that the theory ing degrees of probabilities are invalid and and methodology of biological classification suggest a fuzzy comprehension of scientific must be formulated in terms of the Poppe- philosophy and methodology. rian philosophy of science before a truly After a scientific theory, e.g., a particular valid foundation can be established. classification, has been tested repeatedly Evolutionary classification can be ex- with critical or severe tests and has re- pressed in Popperian concepts—and quite peatedly failed to be disproven, then one easily, I believe, because of the earlier can have confidence in that theory. More analyses of Mayr, Simpson and others. precisely, one says that the theory has been Space does not permit further elaboration corroborated. The term "corroboration" is of details, but a general analysis of the preferred over that of "confirmation" theory of evolutionary classification con- (Popper, 1968b:57) because of the misuse vinces me that it is consistent with Popper's of the latter term and its confusion with basic ideas. r I.C.S.E.B. SYSTEMATIC PHILOSOPHIES 383

A greater problem lies in the method- particular classification already advocated. ology and especially in the tests used in the The fact that an author will not successfully attempts to falsify taxonomic statements. disprove a new classification advocated in Less has been done in this area of evolu- the same publication does not argue against tionary classification and it should there- this sequence of presentation. The ad- fore be an interesting area for future vantage to be gained is that the reader research. Several promising areas of in- knows what statements are available for vestigation can be pointed out. disproof, what tests will be attempted and First is the establishment of which state- hence why certain empirical evidence is Downloaded from ments about affinities are valid scientific being presented. ones and what type of tests can be used in I believe that if attention is given to falsification attempts. In recent years, these methodological problems, that some several authors have argued that statements major gains can be achieved in biological about sister group relationships are testable, classification.

but statements about ancestral-descendent DEFINITIONS http://sysbio.oxfordjournals.org relationships are not testable. Clearly both types of statements are scientific in that The philosophy of definition of words is they are capable of testing and falsification one of the most difficult subjects in philos- by empirical evidence. But the type of tests ophy and one in which I find myself a real and the difficulty involved in attempts to novice. A few points of importance to falsify these two types of statements are biological classification can be made. quite dissimilar. The paramountcy of consistent usage of words cannot be overstressed. Workers in Second is that severe tests for disproving any field of inquiry are constrained to systematic statements must be developed. maintain long established definitions and Not all types of empirical observations pro- usages of words. Otherwise exchange of vide tests of equal severity, and the sorts of ideas becomes difficult if not impossible. at Stanford Medical Center on June 7, 2010 tests that may have been developed in the Numerous needless controversies have attempt to verify taxonomic statements are arisen over the past decade largely because generally radically different from those of redefinition of long established terms that serve to falsify these assertions. Some such as "monophyly" by cladists; I shall use tests are available, as for example the con- these terms in the sense that they have sistency test advocated by Wilson (1965, always had in evolutionary classification. 1967; see Bock, 1969b), the use of par- If the need arises to express new concepts adaptations (Bock, 1967, 1969a), and the in the development of any scholarly pursuit, use of phyletic sequences of derived then new words must be coined. The im- features. portance of maintaining consistent usage of It is interesting to note that phenetic words in discussing classificatory theory approaches appear to provide the best and methodology is as urgent as the need methods to formulate taxa and classifica- to maintain usage in scientific names for tions, but that cladistic methods appear to taxa for which elaborate codes and inter- provide the most severe tests by which to national commissions have been estab- attempt to disprove these taxonomic state- lished. ments. Third is that a change in style of taxo- No philosophy of definitions or of the nomic papers may be useful. Instead of meaning of words is inherently wrong or presenting the classification to be advocated correct. The major impression I have from at the end of the paper among the con- reading a number of articles in this area is clusions, it might be better placed in the the ease with which philosophers can un- introduction. The factual evidence can cover flaws in any general scheme of defi- then be presented in a series of tests de- nition. As such, I reject the notion that signed to attempt the falsification of the only operational definitions are valid in 384 SYSTEMATIC ZOOLOGY science. Operationalism may be a valid methods must be developed by which approach to definition, although the telling objects in nature are recognized and the critique of operationalism by Hempel defined words are applied to them. These (1952, 1965) should be considered care- recognizing methods are analogous to the fully before adopting this approach. It is operations in an operational definition, but my opinion that operational definitions are are separate from the definition and do not poorly suited in a historical biological sci- limit its application within a rigid set of ence such as classification. Thus I prefer bounds. The words, especially in a histori- not to use operational definitions. More- cal science, are often defined in terms of Downloaded from over, it must be emphasized that the failure one set of ideas or concepts and applied to to define classificatory words operationally objects in nature using another set of does not mean that these words cannot be recognizing criteria. Thus the "species" defined. may be defined in terms of reproductive The approach to definition that I favor isolation, but species taxa are generally may be called "theoretical definition" in recognized by means of morphological cri- http://sysbio.oxfordjournals.org which the formal definition of the word is teria (Mayr, 1963). Likewise the criteria the concept and the word, so defined, for definition and those for recognition stands for the concept. Hence, the word differ for "homology" and for "phylogeny." "species" is defined, for example, using the It is not necessary that the criteria used to biological species concept. Other catego- apply the* term to objects in nature be the ries, such as genus, family and so forth, can same as the criteria in the definition. The be equally well defined. The only distinc- two sets of criteria must be interconnected tion between the definition for species and and the use of a particular set of recogniz- those for the higher categories is that the ing criteria must be justified. Moreover, species definition can be associated with a no assurance exists that words will always definite biological phenomenon, namely be applied correctly to objects in nature at Stanford Medical Center on June 7, 2010 lack of gene exchange. Other words, such using any particular set of recognizing cri- as "phylogeny" and "homology," can be teria. Species may be incorrectly described equally well defined in noncircular fashion. and false homologies are often recognized. These definitions, in contrast to operational Such errors testify to the difficulty, and definitions, do not necessarily tell us how the interest, of the science, but do not de- to recognize objects in nature to which the tract from the validity of this approach word can be applied. to definition. A clear distinction must be made be- A final word must be said about cate- tween the word defined and the objects in gories and taxa as the distinction between nature to which the word is applied. In these is still confused by some taxonomists. classification, this distinction exists between Categories, such as species, genus and the word "homology" and particular homo- family, are words and hence are defined. logues that must be recognized. The same Good, clear definitions exist for all cate- distinction exists between categories and gories, although only the species definition various taxa that exist at different cate- can be affixed to a definite biological gorial ranks. Categories such as species, phenomenon. Taxa are groups of organ- genus, and family have been clearly and isms and hence are real objects in nature succinctly defined. Numerous difficulties which are recognized, delimited, described still exist in the methods by which taxa of and named. Taxa are never defined. Names different categorial ranks are recognized are never defined. And it is not correct to and distinguished from one another, but speak of one's concept of a taxon such as these difficulties do not detract from the one's concept of the Dinosauria. These taxa formal definitions. exist in nature, and they are first recog- Words are defined, but then working nized, hopefully correctly, and then de- I.C.S.E.B.—SYSTEMATIC PHILOSOPHIES 385 scribed and named. It may be valid to say use the same philosophical foundation; this that we lack definite, clear and rigid assumption is used in many phenetic ap- criteria with which to recognize and de- proaches to classification. I reject this limit taxa at various categorial levels, and suggestion on the grounds that it is not of that effort should be put into this phase of the greatest use. Comparisons of organisms systematic work. But this is an entirely made under this philosophy do not have different matter from definition of these the greatest possible content. Moreover, categories. Unless this distinction is main- comparisons in all areas of science are not tained clearly, systematics will return to made on the same philosophical basis. Downloaded from the confusion that characterized much of Chemical elements are compared on the the theoretical taxonomic discussions of the basis of the periodic table. Rocks and 1940's and 1950's. minerals may be compared on the basis of their composition and origin, but they PRINCIPLES OF COMPARISON may also be compared on the basis of The philosophy of comparison is closely

coloration which would be less useful in http://sysbio.oxfordjournals.org akin to that of definition and is another petrology. Stars are frequently compared difficult subject. Nevertheless, comparative on the basis of their mass and brightness studies have always been an important part or temperature. of biology, and a bit more can be said than about definition. I do, however, object The best philosophical basis for biologi- strongly to the notion of "the comparative cal comparison, in terms of greatest method" used in a vague sense and to a content, is organic evolution. The rationale sharp distinction between "comparative for this decision is, as stated above, the biology" and "functional biology." Both notion that all attributes of organisms are approaches are and must be used in all the result of their past evolutionary history. areas of biological inquiry. Moreover, the Thus, the most meaningful comparisons are principles and methodology of the com- those made using a set of principles and at Stanford Medical Center on June 7, 2010 parative method, together with the sorts of methodologies based on the theory of interpretations that may be reached validly, organic evolution. This is the philosophy must be stated clearly. I shall comment used, albeit often tacitly, by evolutionary only on a few of the cardinal principles. systematists since Darwin. Much interesting work still remains to be done in formulat- Any comparison made between objects ing and further clarifying the principles of in nature must be based on some definite biological comparison based on evolution- philosophical foundation; the notion that ary theory. Some of these principles may "pure comparisons" can be made is simply be mentioned briefly. invalid. Many different philosophies may exist, both scientific and nonscientific, but 1) The concept of homology is central again none of the scientifically valid ones to all biological comparison and will be may be definitely shown to be correct or discussed separately. wrong. Because the results of any com- 2) Not all comparisons between pairs of parative study are dependent upon the species are the same and not all interpre- philosophical basis accepted, the choice of tations can be applied equally to all com- this foundation is crucial. Herein, I will parisons. A distinction must be made use again the criterion advocated by between horizontal and vertical compari- Popper of higher degree of empirical con- sons (Bock, 1967, 1969a). Horizontal tent or of testability as the basis on which comparisons are those between members to choose a philosophical basis for com- of different phyletic lineages. Vertical parison. comparisons are those between members of the same phyletic lineage. Care must be A commonly offered suggestion is that all exercised in stating that all biological com- comparisons, whether of living organisms parisons are between pairs of species be- or of inanimate objects, are the same and cause a vertical comparison is made be- 386 SYSTEMATIC ZOOLOGY tween different time segments of the same by the genotypic and phenotypic char- phyletic lineage and, as such, cannot be acteristics of the group. Thus the attributes regarded as a comparison between differ- present in members of any taxon should be ent species. A phyletic lineage is formed similar and should be strongly correlated by the consequence of a single species de- with one another. This pattern of similari- scending through time (Bock, 1973). A ties and correlation between features has cross-section of a phyletic lineage at any been the foundation of phenetic methods point in time is a species, but it is not valid that predate Darwin. In recent years, many Downloaded from to consider cross-sections of the same phy- valuable methods have been developed to letic lineage at different times as different analyze these phenetic correlations (e.g., species. Sokal and Sneath, 1963). It must be noted that one often does not 4) In addition to evolutionary change know whether comparisons (actual or being constrained within narrow limits theoretical) are vertical or horizontal be- during the radiation of any taxon, these cause the particular phyletic lineages are modifications occur in a definite sequence. http://sysbio.oxfordjournals.org unknown when the comparisons are made. Any evolutionary modification is dependent Rather, the differing types of interpreta- to a large extent on the preceding changes. tions reached on the basis of these compari- These sequential events can frequently be sons may permit the conclusion of whether reconstructed providing valuable compara- the comparison was horizontal or vertical, tive information that cannot be ascertained and hence whether one is dealing with in phenetic comparisons (e.g., Hennig, members of the same or of differing phy- 1966). These phylogenetic comparative letic lineages. Actual comparisons between methods had their origins in evolutionary all Recent species are, by definition, hori- theory; indeed, Darwin showed the extreme zontal because these species are all mem- value of establishing "pseudophylogenies" bers of different phyletic lineages. as a comparative method. These methods at Stanford Medical Center on June 7, 2010 have been established on a firmer footing Some of the differing interpretations of with the rise of evolutionary morphology horizontal versus vertical comparisons have during the last two decades. been discussed in my earlier papers (Bock, 1959, 1967, 1969a; Bock and de W. Miller, HOMOLOGY 1959). Differences observed in vertical Homology is, without question, the most comparisons are frequently adaptive in important principle in all comparative biol- terms of the selection forces operating dur- ogy. Moreover, it is possible that homology ing the evolutionary changes. Differences is the only method of comparing attributes observed in horizontal comparisons are fre- of different species and that all other quently irrelevant with regard to these methods of comparison are reducible to selection forces. This conclusion has led homology. I do not wish to defend this to the notion of multiple evolutionary path- position in this essay, but would like to ways (Bock, 1959) and to the concept of express the opinion that any comparative paradaptation (Bock, 1967) which are method in conflict with the concept of closely related to the idea of chance versus homology is strongly subject to question design in evolution (Mayr, 1962). The and scrutiny. concept of paradaptation provides the basis A widely accepted definition of "homol- of a valuable test of relationships (Bock, ogy," and the one I advocate, may be 1969a). stated as follows (Bock, 1969a, c): "A 3) The origin and radiation of taxa does feature (or condition of a feature) in one not involve random evolutionary change in organism is homologous to a feature (or the attributes of the group. Rather, only condition of a feature) in another organism some features are modified and these if the two features (or conditions) can be changes are constrained within limits set traced phylogenetically to the same feature I.C.S.E.B.—SYSTEMATIC PHILOSOPHIES 387 or condition in the immediate common an- may be homologous with a more restrictive cestor of both organisms." Thus "homology" conditional phrase. is defined in terms of "phylogeny," and The phylogenetic definition of homology "phylogeny" can be defined in terms of is a nonoperational one and does not pro- "evolution" (Bock and von Wahlert, 1963). vide the methodology by which homologues This definition is noncircular and is not an are recognized. The recognizing criteria operational definition. used in ascertaining homologues are simi- Homology is not an intrinsic property of larities of all sorts (Bock, 1969a:415-416), a feature, such as its mass or color, but is be they of appearance, material composi- Downloaded from a relationship depending upon the existence tion, positional relationship with other of corresponding features in other organ- features, embryological or whatever. More- isms. Because homology is a relational over, similarity between features is the only concept and because the degree of relation- criterion by which homologues can be ship varies, one must always state the recognized. Here I must emphasize that

nature or condition of a particular set of the defining criterion for homology is http://sysbio.oxfordjournals.org homologues. That is, any statement about phylogeny, but the recognizing criterion for the homology of features in different organ- homology is similarity. I do not advocate isms must include a conditional phrase a definition for homology based on describing the nature of the relationship. similarity. Thus it is erroneous to say that the skull The rationale for using similarity as the of a horse is homologous, or to say that the recognizing criterion of homology is simple. humerus of the gorilla is homologous to the If features in different organisms are con- humerus of the chimpanzee. Rather, one sidered to be homologues, then these must say that the wing of birds and the features were the same, and self-identical, wing of bats are homologous as the fore- in the common ancestor. During the course

limb of tetrapods, or that avian wings and of the separate evolutions of the phyletic at Stanford Medical Center on June 7, 2010 chiropteran wings are not homologous as lineages leading to these organisms, the aerodynamic planes. The conditional phrase homologous features in each lineage under- describes the nature of the feature in the went some, and usually different, changes. common ancestor from which the homol- But those attributes of these homologous ogous features stemmed phylogenetically. features that did not modify during the Hierarchies of homologues can be estab- evolution from the common ancestor would lished by restricting the conditional phrase still be the same. Hence shared similarities more and more, and ascertaining whether can be interpreted to be ancestral similari- the features are still homologous. These ties and unchanged since the self-identical hierarchies can be inclusive, nonoverlap- condition in the common ancestor, and ping ones, or they can be sequential ones, hence provide the evidence on which to or a combination of both. The inclusive, judge the homology of features in differ- nonoverlapping hierarchies of homologues, ent organisms. To be sure, similarities thus established, correspond to ever closer can arise via other evolutionary mecha- relationships of the organisms in a phenetic nisms, such as independent origin, paral- sense. The sequential hierarchies, if ar- lelism and convergence (see Throckmorton, ranged according to the presumed phylo- 1965, for an interesting analysis of these genetic series of events, correspond to the complicating factors), which make classi- sequence of derived features in a cladistic fication the difficult and interesting subject sense. it is, but do not affect the statement that similarity between features is the only The use of conditional phrases in homol- method for recognizing homologues. ogy demonstrates that no distinction exists Unfortunately, the methods by which between characters and character states. homologues are recognized and distin- The latter are simply characteristics which guished from nonhomologous features have 388 SYSTEMATIC ZOOLOGY low resolving power (Bock, 1963, 1969a). a valid foundation on which to deduce Thus, numerous errors will be made in classifications can be obtained if several decisions on homologous features, especially hundred features are used. This approach in judging many nonhomologues to be true is based on the assumption that the "noise" homologues. These errors cause problems generated by the incorrect homologues will in actual systematic studies, but they cancel out and not affect the final con- present no difficulties for theoretical con- clusions; this assumption appears to be siderations of the concept of homology. generally valid. Systematists have been far too concerned 2) The careful study of each feature and Downloaded from about making errors in recognizing homo- rejection of those in which the possibility logues. Little can be done because the of error in the judgement of homology is available methodology possesses such low high. The features used in deducing a resolving power and no other methods exist. classification are reduced to a much smaller Decisions on homologues must be made number, but each feature is one in which

and the probable errors may be recognized the taxonomist has more confidence in the http://sysbio.oxfordjournals.org once classifications with a high degree of initial decision on homology. This approach corroboration have been achieved. has been used extensively in evolutionary The sequence of study and decision must classification under the notion of "weighing be carefully observed to avoid the trap of of characters." Much of what has been circular reasoning. The first step is a com- done in weighing of characters is not justi- parative study of features in different fied, but I believe that some of the newer organisms. Decisions on homologues, in- methods developed in evolutionary mor- cluding the conditional phrases, are made phology will permit homologies to be recog- on the basis of observed similarities regard- nized on firmer reasons. My preference is less of the number of errors made. The for this approach mainly because it permits

patterns formed by the homologues, in- many interesting morphological and evolu- at Stanford Medical Center on June 7, 2010 cluding the inclusive, nonoverlapping hier- tionary studies along with the taxonomic archies and the sequential hierarchies of analyses. But I believe that both ap- conditional phrases, are used to deduce proaches, if properly used, provide equally the phylogeny and the classification of the valid bases to deduce classifications. organisms. Needless to say the numerous No information derived from the pre- errors in recognizing homologues causes sumed relationships of a group of organ- great difficulties at this step. And, hence, isms can be used to recognize homologues. the basic principle has been developed of This includes any information from their basing classifications on many characters. classification, phylogeny or possible inter- Two general approaches have been de- mediate position of taxa. Use of such veloped to overcome the difficulty arising information would result in circular reason- from the low resolving power of methods to ing if these homologues were used to test recognize homologues and the many result- the presumed relationships of these organ- ing erroneous homologues. These are: isms. Thus it should be noted (Bock, 1) The use of an extremely large number 1969a:416) that one of Remane's major of characters in deducing a classification as criteria for homology (1956:58) and all advocated by the numerical taxonomists of his supporting criteria depend on earlier who recognized this problem clearly. They conclusions about relationships and should argue, perhaps tacitly, that many of the not be used. Theoretical analyses of com- homologues used in classification are parative methods (e.g., Bock, 1963, 1969a) wrong, but the use of a large number of in which "false homologies" are discussed characters will provide sufficient correct on the basis of classifications with a high homologues, even if they are unknown, on degree of corroboration must be done with which to base a classification. Even if only care because the "false homologies" may ten percent of the homologues are correct, I.C.S.E.B.—SYSTEMATIC PHILOSOPHIES 389

not be incorrect at all and may represent homologues does one have the greatest the critical test disproving the classification. confidence. All statements about homologous features The statement that all valid methods of and phylogenies are scientific theories, and classification must be reducible to the con- hence may be disproven, but never verified. cept of homology, or at least not be in con- At best they can acquire a high degree of flict with it, suggests a serious internal con- corroboration, after which systematists have flict in cladistics as formulated by Hennig such confidence in them that they rarely (1966). To my knowledge, no later pro- bother to test them. ponents of cladism have rectified this con- Downloaded from All comparative methods in biology de- tradiction. The conflict stems from the pend, I believe, on homology. The length above assertion that all comparative of time involved in the comparative study methods are based on the concept of of features, deciding homologies and de- homology—that homologues must be ducing classifications is of no importance. recognized before further work is possible

The initial sorting of specimens into species —and that the only criteria for recognizing http://sysbio.oxfordjournals.org and grouping of these species into taxa of homologues are similarities of all sorts. This higher rank may be done extremely rapidly must be coupled with the notion that on the basis of superficial examination. homology is a relative concept and that all Likewise, the recognization of new species statements about homologues must contain and their assignment to the correct higher a conditional clause giving the nature of taxa may be done very rapidly by an the homology. experienced taxonomist. But the procedure Hennig (1966:10; 12; 23; and 88) states followed, no matter how fast it is done, is clearly that similarity is not a factor used to make a large number of decisions about in judging relationships; no possibility ex- homologues complete with conditional ists to interpret his statements otherwise. phrases based on recognizing criteria of Relationships are based on common posses- at Stanford Medical Center on June 7, 2010 similarity and then to use these homologues sion of derived homologous features. Yet, to deduce relationships. The speed at which the human mind can operate should he (1966:93-94) accepts Remane's (1956) not be underestimated, even compared to definition and methodology of homology modern high-speed computers. But this although this part of Hennig's book is speed should not be misinterpreted as some rather confused. Remane's methods of other procedure used in classification. recognizing homologues are basically the Taxonomists do not establish taxa and same as those outlined above; they depend systems of classification by some method on similarities. Thus it is not possible to other than homology and then test them by reject similarity between features as a studying homologies. (One can, of course, factor in deducing classification and to establish classifications by some stochastic base classification on homologous features method and then test them, but few taxon- which is a primary requirement of any omists would be interested in this ap- approach to biological classification. This proach.) Classifications are deduced on contradiction is a serious one, and unless the basis of previously established homo- it can be resolved by cladists, I believe that logues, no matter how rapidly done, which it eliminates the distinction claimed by then serve as falsification tests for the cladists between their approach to classi- classification. More severe falsification tests fication and that of evolutionary taxon- can be undertaken by careful study of omists. certain features, which is the meaning of As a conclusion, I would like to offer the statements that taxa are established first opinion that any approach to biological and then tested by studying homologues. classification that excludes homology and In such tests, one must always decide which its recognizing criterion of similarity as a homologues are "best," that is, in which primary step in deducing relationships is invalid. 390 SYSTEMATIC ZOOLOGY

WEIGHING AND TESTING features. Such a procedure is not satisfying The concepts of weighing of characters though to many workers. and testing of classifications appear to have The major concern is that circular reason- little to do with one another, but develop- ing must be avoided in assigning relative ment of proper tests of falsification have weight. Features must be weighed before considerable bearing on the importance of the taxa are recognized or the classification weighing of characters. The whole subject deduced. Thus arguments that features of relative weight of taxonomic characters unique to a group should be given more is riddled with difficulties, has been weakly weight are invalid. So are conclusions that Downloaded from defended against major challenges, and yet conservative features be weighed more has emerged unscathed. Satisfying argu- unless independent evaluation of conserva- ments supporting the notion of weighing tive characters can be given. High cor- characters are hard to find, but really relation with other features is probably not devastating arguments against this idea do a valid criterion for assigning great weight

not exist. because this decision is dependent upon http://sysbio.oxfordjournals.org Only a slight experience with actual prior recognization of taxonomic groups. taxonomic problems is needed to realize Some valid criteria do exist for judging that different features possess varying de- relative value of taxonomic characters. grees of usefulness for classification, and Perhaps the best and most widely used is that some form of weighing characters is the complexity of the feature. A complex useful. The alternative is not to weigh attribute with many interconnected parts characters and to use very large numbers, is usually given more taxonomic weight as done in some forms of phenetic analyses. than a simpler one because the possibility Although the basic assumptions of non- of independent evolutionary origins of a weighing approaches are valid, no one has complex feature with many similar parts is

really solved two problems. The first is the less than the chances of independent evo- at Stanford Medical Center on June 7, 2010 assumption that the noise of the large lution of simple features. The evolutionary number of "poor" characters really cancels origin of a new feature or major change in out without affecting the classification. an existing feature is usually given greater Second is whether the efforts of measuring, weight than the loss of a feature, again be- recording and otherwise handling the large cause the chances of independent loss of a number of characters needed in nonweigh- feature is greater than the independent ing approaches are justified by the results origin of similar features. Most of the suc- obtained. cessful applications of character weighing The major objection against weighing of have used criteria such as these, but prob- characters is that few workers ever suggest lems arise again when features of similar what they mean by better characters or out- complexity suggest conflicting classifica- line the procedures used to weigh them. tion. Opponents to weighing are probably justi- The whole notion of character weighing fied in claims that most weighing is done may be of considerably less importance to a posteriori after the classification had been classification than believed earlier once the established. full implications of the Popperian phi- Nevertheless, if one accepts the validity losophy are applied to taxonomy. If of evolutionary phenomena of independent classifications and phylogenies are deduced origin and convergence of features, and statements available for disproof by empiri- realizes that many decisions about homo- cal testing, a very different approach to logues are wrong, then it must follow that taxonomic features can be used in which some features are better clues to relation- relative weight may have no role. If some ships and should be given more weight. empirical evidence disproves a statement This can be done even if one cannot state of relationship, then it matters little clearly why more weight is given to some whether the features are of great value or I.C.S.E.B.—SYSTEMATIC PHILOSOPHIES 391 of little value taxonomically. More essential Because in evolutionary classification, the is the development of severe or critical tests phylogeny and the formal classification are which can provide a strong basis of falsifi- not different but redundant images of each cation. Such tests, as Wilson's consistency other, neither one by itself is sufficient. To tests, do exist, as mentioned above, but be most useful, the results of any study of more effort is needed to develop and evalu- the relationships within a group of organ- ate them. isms must include a formal classification It should be obvious to working system- and a phylogenetic diagram in its con- atists, that such tests will be no better than clusions. Both are needed to provide the Downloaded from the characters used in them. And appli- maximum information about relationships cation of falsification tests will soon result required by other biologists for their com- in the situation of two severe tests using parative studies. different features with conflicting results. Such situations will lead right back to CONCLUSION evaluation of characters and hence argu- In conclusion, I believe that a firm http://sysbio.oxfordjournals.org ments of weighing. Before we worry about philosophical foundation exists for classi- completing the circle, I believe that serious cal evolutionary classification and can be attempts should be made to develop a good expressed in the form of a set of princi- series of falsification tests, apply them and ples. And I believe that classical evolu- assess the results before pursuing further tionary classification is the best approach arguments on weighing. to biological classification which must pro- vide the foundation for all comparative CLASSIFICATION AND PHYLOGENY studies in biology. Classical evolutionary The only question that I would like to classification is based on the Popperian investigate is whether the phylogeny of a philosophy of scientific demarcation and group and the formal classification for that methodology and on the modern synthesis at Stanford Medical Center on June 7, 2010 group must be absolutely concurrent. The of Darwinian evolution—namely that clas- principle that the formal classification sification of organisms is based on their should reflect the evolution of a group does entire evolutionary history not just on their not mean that a one to one relationship phylogeny or on their degree of similarity. must exist between the classification and In this eclectic approach, relationship be- the phylogeny of the group. Nor is such a tween organisms, as expressed in formal relationship necessary. The basic principle taxa, attempts to maximize simultaneously of evolutionary classification, and I believe the two semi-independent variables of de- most biological classification since Darwin, gree of similarity and of phylogenetic is that classification reflects the evolu- sequence of events. The resulting classifi- tion, not just the phylogeny, of the group. cation may be subjective in that each tax- And under the concepts of evolutionary onomist will not necessarily reach the same classification, in which the formal classifi- conclusions, but it provides the most cation is an attempt to maximize simultane- realistic picture of the complexities of ously the two semi-independent variables biological diversity. of genetic similarity and phylogenetic Lastly, although the focus of this essay sequence, a one to one correlation between is on the philosophical foundation of the classification and the phylogeny of the biological classification and on theoretical group is impossible. considerations of methodology, the major Thus it is not possible to provide simple plea should not be for emphasis in this rules of how to derive the classification phase of systematics but for more detailed from the phylogeny of a group, or of how systematic analyses of plant and animal to regain the same phylogeny from the groups. Historical study of the improve- classification. ment in the classification of any taxon 392 SYSTEMATIC ZOOLOGY suggests that the major reason for improved BOCK, W. J., AND G. VON WAHLERT. 1963. Two comprehension of the systematic relation- evolutionary theories—a discussion. Brit. J. Phil. Sci. 14:140-146. ships between these organisms has been a COLLESS, D. H. 1967. The phylogenetic fallacy. more thorough, detailed study of their Syst. Zool. 16:289-295. characteristics, not the application of new GHISELIN, M. T. 1969a. The triumph of the or different philosophical approaches to Darwinian method. Univ. Calif. Press, Berkeley classification. Continued success and sur- and Los Angeles, X + 287 pp. vival of classification as a biological science GHISELIN, M. T. 1969b. The principles and

concepts of systematic biology. In: Systematic Downloaded from will depend on the abilities of systematists Biology. Nat. Acad. Sci., publ. 1962:45-55. to produce improved classifications that HEMPEL, C. G. 1952. Fundamentals of con- serve as the basis for all comparative studies cept formation in empirical sciences. Univ. of in biology. Chicago Press, Chicago. HEMPEL, C. G. 1965. A logical appraisal of ACKNOWLEDGMENTS operationism. In Aspects of Scientific Expla- nation. The Free Press, New York.

My understanding of the topics covered HENNIG, W. 1966. Phylogenetic systematics. http://sysbio.oxfordjournals.org in this essay have been improved greatly as Univ. Illinois Press, Urbana, 263 pp. a result of discussions with a number of MAYR, E. 1962. Accident or design, the paradox of evolution., pp. 1-14. In: G. W. Leeper (ed.) friends; I wish to express my gratitude to The evolution of living organisms. Melbourne all and especially to Ernst Mayr, Bobb Univ. Press, Victoria, Australia. Schaeffer, Lester Short, F. E. Warburton MAYR, E. 1963. Animal species and evolution. and Gerd von Wahlert who have borne the Harvard Univ. Press, Cambridge, Mass., xvi -f- brunt of these talks. The morphological- 797 pp. MAYR, E. 1969. Principles of systematic zool- systematic research, on which my theoreti- ogy. McGraw-Hill, New York, 428 pp. cal deliberations are based, were done with POPPER, K. 1968a. The logic of scientific dis- the support of several grants from the covery. Harper Torchbooks, New York and

National Science Foundation. This manu- Evanston, 2nd Edition, 480 pp. (originally at Stanford Medical Center on June 7, 2010 script was written during the tenure of published as Logik der Forschung, 1934, first Grant GB-37305 X from the National English edition, 1959). POPPER, K. 1968b. Conjectures and refutations: Science Foundation. The growth of scientific knowledge. Harper Torchbooks, New York and Evanston, xiii + 417 REFERENCES pp. (originally published, 1962, Basic Books, N. Y.). BOCK, W. J. 1959. Preadaptation and multiple evolutionary pathways. Evolution 13:194-211. POPPER, K. 1972. Objective knowledge: An evolutionary approach. Oxford Univ. Press, BOCK, W. J. 1963. Evolution and phylogeny in morphologically uniform groups. Amer. Natur. Oxford, x + 380 pp. 97:265-285. REMANE, A. 1956. Die Grundlagen des natur- BOCK, W. J. 1967. The use of adaptive char- lichen Systems, der vergleichenden Anatomie acters in avian classification. Proc. XIV Internat. und der Phylogenetik. Akad. Verlagsgesell., Ornith. Cong., pp. 61-74. Leipzig, 364 pp. BOCK, W. J. 1969a. Comparative morphology SIMPSON, G. G. 1961. Principles of animal in systematics. Syst. Biology, Nat. Acad. Sci., taxonomy. Columbia Univ. Press, New York, publ. 1962:411^48. xii + 247 pp. BOCK, W. J. 1969b. Nonvalidity of the "phylo- SOKAL, R. R., AND P. H. A. SNEATH. 1963. genetic fallacy." Syst. Zool. 18:111-115. Principles of numerical taxonomy. Freeman & BOCK, W. J. 1969c. The concept of homology. Co., San Francisco and London, 359 pp. Ann. N. Y. Acad. Sci. 167:71-73. THROCKMORTON, L. H. 1965. Similarity versus BOCK, W. J. 1973. Examination of macroevolu- relationship in Drosophila. Syst. Zool. 14:221- tionary events by sequential species analysis. 236. Proc. 17th Internat. Zool. Cong. 17 pp. WILSON, E. O. 1965. A consistency test for BOCK, W. J., AND W. DE W. MILLER. 1959. The phylogenies based on contemporaneous species. scansorial foot of the woodpeckers, with com- Syst. Zool. 14:214-220. ments on the evolution of perching and climbing WILSON, E. O. 1967. The validity of the "con- feet in birds. Amer. Mus. Novitates, No. 1931: sistency test" for phylogenetic hypothesis. Syst. 1-45. Zool. 16:104. I