University of Virginia Library

BIOLOGICAL HOMOLOGIES
AND ANALOGIES

Introduction. In his study of evolution, or arrange-
ment of organisms into taxa, the systematic biologist
makes continual use of the concepts of homology and
analogy. The historical development and current
meanings of these concepts are of interest both because
they help show how the science of systematics is done,
and because of the intriguing frequency with which
philosophically mistaken arguments have intruded into
the science.

The fundamental aim of taxonomy is to classify
organisms into groups in a biologically meaningful way.
The Linnean hierarchy, built along lines formally pro-
posed by Aristotle, is the familiar result. A major pur-
pose of evolutionary biology is to discover the actual
diverging sequences of organisms, commonly described
by phylogenetic trees, and the causes of those diver-
gences.

Historical Development of the Concept of Homol-
ogy. It is in the light of these goals and difficulties
that one should consider the prevalent evolutionary
definitions of homology and analogy. According to
G. R. De Beer, “The sole condition which organs must
fulfill to be homologous is to be descended from one
and the same representative in a common ancestor.”
G. G. Simpson also defines homology in terms of com-
mon inheritance: “Homology is resemblance due to
inheritance from a common ancestry.” Usually, this
evolutionary sense of homology is contrasted with
analogy, resemblance due to common function. A fre-
quent example of homology is the relation between
a bird's wing and a mammalian forearm; of analogy,
the relation between the wing of a bird and of an
insect.

It has recently been claimed by several authors that
the evolutionary definition of homology is viciously
circular, for “in order to show that a part of one
organism, x, is homologous with a part y of another
organism, it must be shown that they are derived from
a part z in a common ancestor. But homology itself
is invoked in identifying parts x and y with part z”
(Jardine, 1967). It will be argued below that the circu-
larity of De Beer and Simpsons' definition is not at
all vicious, but whether it is or not, the circularity
points up the priority of a nonevolutionary, phenotypic
sense of homology, defined solely in terms of pheno-
typic similarity, and logically independent of criteria
of descent.

A nonevolutionary sense of phenotypic homology
was widely utilized prior to Darwin. In 1818, Geoffrey

St. Hilaire, in his théorie des analogues (where analogue
is roughly equivalent to phenotypic homology), argued
that animals conform to a common type; the analogy
of parts was to be established by showing that they
occupied corresponding relative positions in different
animals—the principe des connexions. Owen, in his
Lectures... (1843), defined homologue as “The same
organ in different animals under every variety of form
and function.” In 1847 he distinguished special
homology “the correspondence of a part or organ,
determined by its relative position and connections,
with a part or organ in a different animal,” from gen-
eral homology, “a relation in which a part or a series
of parts stands to the fundamental or general type.”
Several authors in addition to Owen, most notably the
poet Goethe, advocated this Platonic idealism of fun-
damental types, or common bauplans (construction
plans) of which diverse organisms were seen as imper-
fect realizations.

Darwin himself utilized Geoffroy St. Hilaire's con-
cept of homology in the Origin of Species... (1859):
“... If we suppose that the ancient progenitor, the
archetype as it may be called, of all mammals, had
its limbs constructed on the existing general pattern,
for whatever purposes they served, we can at once
perceive the plain signification of the homologous
construction of limbs throughout the whole class.”
Darwin, of course, thought homology evidence in favor
of evolution, but he utilized a nonevolutionary, pheno-
typic concept of homology.

Both the Platonic idealism of fundamental types, and
the nonevolutionary sense of homology soon came
under attack. In 1870,  E . R. Lankester argued that St.
Hilaire's and Owen's use of homology “belongs to the
Platonic school.... Professor Owen... would under-
stand by homologue 'the same organ in different ani-
mals under every variety of form and function'....
But how can the sameness (if we may use the word)
of an organ under every variety of form and function
be established or investigated?... to settle this ques-
tion of sameness, an ideal 'type' or a group... had
to be evolved from the human mind,... then it could
be asserted that organs might be said to be the 'same'
in two animals” (“On the Use...,” pp. 34-43). In the
place of this idealist, purely phenotypic concept of
homology, Lankester proposed a definition resting in
part on common ancestry. “Structures which are ge-
netically related, in so far as they have a single rep-
resentative in a common ancestor, may be called
homogenous” (idem). The term “homogeny” was not
accepted, but its definition was used to define
“homology” in the evolutionary sense later advocated
by Simpson, De Beer, and others.

Are Lankester's points well taken? It is true that
Owen's and Goethe's fundamental type or common
bauplan has the features of a Platonic Ideal of which
all organisms of the type were imperfect realizations.
Their major motivation for the supposition of a funda-
mental type was the belief that in order to class a group
of organisms together, the organisms must jointly share
some common features, the fundamental type.

This understanding of the nature of universals has
been criticized by the philosophers Ludwig Wittgen-
stein and (more recently) Morton Beckner. Wittgen-
stein points out that we apply the concept “game” to
diverse objects which share different attributes in par-
tially overlapping ways but have no single attribute
in common. Classification of objects together requires
only that we have repeatable criteria for their classifi-
cation. Thus phenotypic homology need not lead to
idealism.

Furthermore Lankester's definition of homogeny
(homology) failed to meet his own criticisms of Owen,
for to establish that two structures have a single repre-
sentative in a common ancestor requires showing that
the two structures are homologous to that single an-
cestral representative; and the only criterion of homol-
ogy to which Owen can allude is phenotypic homology.
If phenotypic homology inevitably led to idealism, then
Owen's redefinition of homology would fail to escape
the disaster.

There were, however, more valid grounds for the
introduction of Lankester's sense of homology. It was
argued that evolutionary theory provided a criterion
by which to distinguish “true” from accidental homol-
ogy. Because the traits of organisms overlap in complex
ways, phenotypic clusters can be made discrete only
somewhat arbitrarily. Hence, phenotypic homology is
somewhat arbitrary. However, evolution of organs in
one species took place, presumably, in only one way.
By utilizing the criterion of descent of two structures
from the same representative in a common ancestor,
evolutionists like Simpson and De Beer felt they were
providing a less arbitrary and more biologically mean-
ingful sense of homology. With the wide adoption of
this evolutionary sense of homology, the meaning of
the term had changed. In its new sense, homology
required the theory of evolution for its definition. This
is one of many cases in science in which a theory built
in part with data described by one concept, later is
used to change the meaning of that concept itself.

But the clarity of aim of the evolutionists to intro-
duce a concept of homology which overcame the ar-
bitrariness of mere phenotypic homology does not
completely attain its goal. To establish that two struc-
tures are homologous, in Simpson's sense, requires

showing that a phylogenetic hypothesis is true. But part
of the evidence that the hypothesis is true stems from
phenotypic resemblance of structures in fossil and
living organisms. And that phenotypic homology
suffers from the arbitrariness which the evolutionist
hoped to avoid. Nevertheless what the evolutionist does
do is to construct and attempt to verify phylogenetic
hypotheses on the basis of all the available information,
including not only phenotypic homology but also tem-
poral relations among structures in the fossil record,
as well as considerations about possible sequences of
change in structures, which minimize the total number
of changes which must be supposed to have occurred.
For the evolutionist, the statement that two structures
are homologous is the result of a great deal of theory
building. Because the evolutionist utilizes other criteria
in addition to phenotypic similarity in asserting that
two structures are homologous, he is able to say that
two phenotypically dissimilar structures are homol-
ogous; for example, the bones of the mammalian mid-
dle ear and their homologues in fish.

The evolutionary sense of homology has held sway
for nearly a century. It is currently facing vigorous
attack by systematists who wish to substitute a purely
phenotypic sense of homology. The argument is not
merely semantic, for it expresses very different con-
victions about how best to do systematics.

The pheneticists (who wish to classify organisms on
the basis of clusters of phenotypic traits) raise three
major arguments against the traditional evolutionary
taxonomist. 1. The evolutionary sense of homology is
claimed to be viciously circular, for homology must
be utilized to show that two structures are descended
from the same representative in an ancestor. 2. The
pheneticists argue that the traditional evolutionist has
poorly defined criteria for asserting the phenotypic
similarity of organisms. 3. Worse, it is argued, the
traditional taxonomist interjects speculative phylo-
genetic hypotheses into his very phenotypic classifica-
tion schemes, thus rendering them biased and unscien-
tific, and rendering suspect any purported evolutionary
homologies derived from the data.

In place of these putatively objectionable practices,
the pheneticist, and in particular, the numerical tax-
onomist, wishes to substitute less biased, more reliable
methods (Sokal and Sneath, 1963); therefore, they have
introduced “operational homology” (Sokal and Camin,
1965). Unit characters, such as eyes: red, blue, green,
etc., are chosen (Colless, 1967); and, on the basis of
a set of such characters and their “states” for each
organism in a sample, similarity among groups of orga-
nisms is computed. Many techniques of calculation of
similarity have been generated (Sokal and Sneath),
depending upon whether each unit character is con
sidered of equal or differing weight. Diverse functions
on these characters are computed as measures of simi-
larity or “distance” between organisms, and diverse
types of cluster and factor analysis are utilized to
generate clusters.

The claims for scientific respectability of these tech-
niques are: 1. The criteria by which organisms are
judged similar are explicit and repeatable, in contrast,
the pheneticists allege, to the unclear criteria by which
traditional taxonomists talk of similarity among orga-
nisms. 2. The phenotypic clusters found by these
techniques are supposed to be free of phylogenetic
speculations, or, more strongly, free of theory, and
unvarnished data on which to build theory.

Can these arguments be maintained by the pheneti-
cist against the traditional evolutionary taxonomist?
First, the circularity ascribed to the evolutionary sense
of homology also applies to the purely phenotypic
sense of homology. The evolutionist's homology is
circular because of his reference to the “same” an-
cestral structure; but that “same” is just the circle the
phenotypic sense of homology requires. If it is a vicious
circle, both evolutionist and pheneticist would be
trapped, but the circle is benign. Many concepts can
only be defined by reference to a self-defining set of
terms. For example, “homologus,” “similar,” “resem-
bling,” “almost the same,”... etc. Benign circles are
not limited to biological concepts, a “rule” cannot
be understood without the ideas of “correct” and
“wrong,” nor they without it.

The pheneticist's hope for a “theory-free” opera-
tional homology is illusory. No operation is completely
free of theory; measuring the same ear length again
requires a theory about measuring rods not changing
length in these circumstances. Nor are the pheneticists'
similarity measures free of phylogenetic bias. Different
choices of unit characters or different computation
schemes will yield different phenetic clusters consistent
with diverse phylogenetic sequences. Suppose the
pheneticist deduced from his clusters a presumptive
phylogenetic sequence which happened to be incon-
sistent with the temporal relations in the fossil record.
Faced with disconfirming evidence, he might doubt the
deductions, doubt the fossil record, or doubt the ade-
quacy of his clustering techniques or choice of unit
characters. If he were willing to consider altering his
unit characters or clustering techniques, then he would
be doing what he accuses the traditional evolutionist
taxonomist of doing, namely, redefining his phenotypic
classification scheme to fit with other phylogenetic data
or hypotheses. If he would never change his clustering
technique, he might be asked to justify his position,
and would find it hard to do so.

Finally, the pheneticist is wrong in asserting that

the traditional evolutionist taxonomist must be unsci-
entific since he cannot make explicit and simple the
criteria by which he judges organisms similar. An art
critic can correctly recognize a Picasso, but would
probably be unable to make all his criteria explicit and
simply measurable. While the numerical taxonomist's
measures are probably more repeatable, and perhaps
more easily learned than those of the traditional tax-
onomist, numerical methods nevertheless are not nec-
essarily more meaningful measures of phenotypic simi-
larity than the traditional evolutionist methods.

In brief, the differences between those who support
a phenotypic and those who favor the evolutionary
sense of homology is not as great as the furor of current
debate makes it appear. Both utilize a phenotypic sense
of homology. The pheneticist must admit that his oper-
ational homology is neither theory-free nor unbiased
with regard to phylogenetic hypotheses, and that the
evolutionary taxonomist is not necessarily unscientific
for his lack of simple measuring operations.

Serial, Sexual, and Genetic Homology. While the
evolutionary and phenotypic senses of homology are
the most fundamental, several derivative senses of the
term are commonly utilized. Serial, sexual, and genetic
homology apply to entire organs; amino acid sequences
in proteins and base sequences in DNA allow the
application of homology at the molecular level.

Serial homology refers to more or less identical,
repeating structures in an organism, for example, the
vertebrae or teeth. The criteria for serial homology
include phenotypic homology of different structures
in the same adult, phenotypic homology of structures
in the embryo, similarity of connections of the repeat-
ing parts, and phylogenetic arguments. The phenotypic
similarity of teeth, vertebrae, ribs, etc. of an adult
mammal are obvious. Even if vertebrae in the adult
were grossly different, however, their claim to serial
homology might still be made on the basis of the serial
homology of the embryonic somites from which they
arose. Thus, ontogenetic data can be utilized to dem-
onstrate serial homology. Similarity of connections of
the humerus and femur help establish them as serially
homologous structures. Even if structures were very
different, if they could be shown to have evolved from
structures which were themselves serially homologous,
then the derived structures might also be said to exhibit
serial homology.

The occurrence of serially homologous structures
provides an important clue about both evolution and
ontogeny. It appears that it is relatively easy to evolve
by changing the number of repeating units which occur
in an organism. Unfortunately, very little is known
about how repeating structures are generated, or how
their number is controlled.

Sexual homology refers to structures which differ
between the two sexes of the same species, but which
derive from a common embryonic rudiment. For ex-
ample, the mammalian penis and clitoris are homolo-
gous structures deriving from the same region of the
genital ridge. The criteria for sexual homology are
therefore the phenotypic homology of embryonic parts
and careful comparative anatomy of the stages of em-
bryonic development in the two sexes.

In its original sense, genetic homology meant that
if structures in two organisms were the consequence
of the action of the “same” gene in both organisms,
then those structures were homologous. The homology
of the structures was a consequence of the homology
of the gene(s). Unfortunately, there is no simple corre-
spondence between the genotype and phenotype of an
organism. Alteration of a single gene may have effects
on many phenotypic traits, and alteration of each of
many genes may have the same effect on a given
phenotypic character. A consequence of this com-
plexity is the great difficulty in trying to prove that
a given phenotypic character in two organisms is due
to the action of the “same” gene(s) in both. It is the
common experience of geneticists that if two initial
populations undergo an identical selection regimen
which successfully maximizes some trait, then the gene
modifications which underlie that change in the two
populations can differ, and usually do differ strikingly.
Application of the notion of genetic homology in such
instances is useless. Because of these difficulties, genetic
homology is not widely utilized.

Ontogenetic criteria of homology had wide appli-
cation when the Recapitulation Theory was accepted.
If ontogeny recapitulates phylogeny, then to establish
the evolutionary sense of homology between two or-
gans—that they were derived from the same part in
a common ancestor—it was sufficient to show they had
the same origin in ontogeny. In 1870, Karl Gegenbaur
stated that special homology is the relationship be-
tween two organs which have had a common evolu-
tionary origin, and which, as a corollary, have arisen
from the same embryonic Anlage. However,  E . B.
Wilson (1895, p. 101) and De Beer (1958) have shown
cases of homologous adult structures derived from
different embryological origins. Such a straightforward
application of ontogenetic criteria no longer suffices
to prove that two adult structures are homologous in
the evolutionary sense.

Molecular Homology. In the 1950's and 1960's the
notion of homology has begun to be extended to the
molecular level, to the comparison of nucleotide se-
quences in the DNA's of different organisms, and to
the comparison of amino acid sequences in proteins
from different organisms. If the work to establish mo-

lecular homology is great, so too are the conceptual
rewards. One of the most difficult handicaps of classical
genetics and evolutionary studies is, as we have noted,
the lack of a simple relation between the genotype
and phenotype of an organism. A consequence of this
complexity is that if two classes of organisms exhibit
a certain degree of phenotypic difference, one cannot
usually determine the extent of genotypic difference
between the two classes. Thus, from classical genetics
we can usually know neither the extent of genotypic
change underlying the observed phenotypic alterations
in evolution, nor, therefore, the rate at which the
genotype changes in evolution.

These difficulties are partially overcome by consid-
ering nucleotide sequences in DNA and amino acid
sequences in proteins. Since the nucleotide sequence
of the DNA is the genotype, comparison of nucleotide
sequences in different organisms is the most direct
means of assessing the extent of genetic change in
evolution, and genetic homology between species.
However, such an assessment is not as straightforward
as one might have hoped. In the first place, direct
analyses of truly long sequences of bases in DNA are
not currently available. Estimations of similarity of
nucleotide sequences between DNA's utilize indirect
techniques which only establish approximate ho-
mology, not identity of sequence. These techniques
will be described later.

Even were it possible to obtain quite detailed nu-
cleotide sequences for the DNA of two organisms, say
bacteria, estimation of the extent of genotypic differ-
ence between the two would remain difficult. The
concept of the extent of genotypic difference is ambig-
uous. Ambiguity resides in the dual reference of
“genotype” to the actual physical structure of the
DNA, the sequence of bases, and also to the DNA as
the carrier of genetic information. If one is referring
to the physical genotype, then the extent of difference
between two genotypes is simply the number of ho-
mologous loci at which the nucleotides differ. Alter-
ation in the informational genotype is related in rather
complex ways to alteration in the physical genotype.
Amino acids are coded for by triplets of nucleotides;
most amino acids are coded for by two or more codons.
Thus, some nucleotide substitutions change a codon to
a second codon for the same amino acid. Such a substi-
tution alters the physical genome, but leaves the infor-
mational genotype unaltered. Physical genotypes
different at many loci can be the same informational
genotype. Conversely, nearly identical physical geno-
types can be radically different informational geno-
types. This possibility is a consequence of the fact that
codons are triplets, and an amino acid sequence is
specified by a sequence of triplets in which the nucleo
tides are “read” from a specific starting point, three
at a time. A deletion of a single nucleotide can cause
a “reading frame shift” in which all codons downstream
from the deletion are misread and a large number of
incorrect amino acids are incorporated into the pro-
tein. A small change in the physical genotype yields
a large change in the informational genotype. If one
is concerned with the extent and rate of alteration of
the informational genotype in evolution, one must view
with caution data derived from estimates of physical
homology of the DNA's of various organisms.

Indirect physical techniques to study the extent of
base sequence homology of the physical genotype de-
pend upon the DNA's duplex structure whose comple-
mentary strands may be separated and caused to re-
combine. Since single-stranded DNA components from
different origins may also be induced to form “hybrid”
structures, a means is afforded by which to assess ge-
netic relationships among organisms. It can be shown
that duplex formation between single strands derived
from DNA of the same or nearly identical species
occurs readily, but fails to occur if the strands are
derived from very different organisms.

Results of such studies (Bolton, p. 77) indicate that
phenotypically similar animals have very similar DNA
base sequences. Furthermore, “... the similarities and
differences in polynucleotide sequences quantitatively
indicate the extent of the taxonomic category to which
the systematist refers. Thus, among the primates, a
superfamily distinction means that about one-quarter
of the polynucleotide sequences are different, half are
different for subordinal separation, and about three-
quarters for ordinal distinction.” Bolton also notes that
“the quantitative similarities in polynucleotide se-
quences among vertebrates can be related to the time
at which the lines of organisms in the present diverged
from one another in the geologic past according to
the paleontologist's judgment.” Bolton's figure shows
a linear decrease in the logarithm of DNA similarity
with time.

While Bolton's data gives a good indication of the
rate of alteration of a physical genome in evolution,
it remains difficult to relate the results to the extent
and rate of change of the informational genotype in
evolution.

A conclusion reached by Britten and co-workers
(1968, p. 529), is that many nucleotide sequences occur
repeatedly in the DNA of higher organisms, there being
many DNA families, each with many nearly identical
copies of one sequence. The existence of these homolo-
gous DNA sequences renders the relation between the
physical and informational genotype even more com-
plex, for the functional significance of the redundant
DNA is not known. Britten's data also casts doubt on

Bolton's conclusion about DNA homology among spe-
cies, for Bolton probably measured only highly re-
dundant DNA sequences.

In contrast to changes in nucleotide sequences which
may occur without alteration of the informational
genotype, changes in amino acid sequence are evi-
dence, by definition, of alteration of the informational
genotype. With the exception of substitutions of nu-
cleotides which do not change the amino acid specified,
substitution of a single nucleotide results in the substi-
tution of a single amino acid at a locus in the poly-
peptide. Since the assignment of codons to amino acids
is now fairly well established, it is now possible to say
which amino acid substitutions can occur by substi-
tution of a single nucleotide in a codon. Some amino
acid substitutions cannot be made by altering a single
nucleotide, but would require the simultaneous alter-
ation of two or three nucleotides; or else, since nucleo-
tide substitutions must usually occur one at a time,
intermediate proteins with an amino acid different
from both the first and final form, must have existed.

Partial or complete sequences of amino acids have
now been worked out for several sets of homologous
proteins in different organisms, for example, hemo-
globin and cytochrome C (Fitch and Margoliash, 1967).
By utilizing arguments about minimal possible changes
causing sequences of amino acid substitutions, coupled
with assumptions about nonreversal of changes, it is
possible to arrange contemporary proteins into pre-
sumptive branching phylogenetic sequences (ibid.).

Evidence supporting the deduced branching phylo-
genetic relations can be sought in the fossil record. The
form of argument utilized is closely similar to that
noted by  E . O. Wilson in 1965 for deducing consistent
possible phylogenies based on gross phenotypes of
contemporary organisms. Utilizing such techniques on
cytochrome C, Fitch and Margoliash (1967) have pro-
duced a phylogenetic tree linking fungi, yeasts, nema-
todes, fish, birds, and mammals, which is very similar
to phylogenetic trees proposed by classical zoologists.
The number of amino acid substitutions, coupled with
time estimates derived from the paleontological record,
can give an estimate of the rate of mutation of the
informational genotype. It will be of particular interest
to compare the rates for proteins performing diverse
functions, for the rate must depend in part upon the
strictness of selective constraints on workable amino
acid sequences.

The extent of homology in amino acid sequence for
some proteins is enormous; neurohypophysial peptide
hormones hardly differ from man to shark (Acher,
1969). Other proteins exhibit far less homology, differ-
ing in many loci in many different ways. The occur-
rence of such proteins, all performing the same func
tion in different animals, has led some biologists to
suppose that some amino acid substitutions do not
affect protein function and are therefore not subject
to selection. By random drift, large numbers of such
substitutions are claimed to accumulate, so that these
homologous proteins differ at many loci but continue
to function.

Amino acid sequence homology is also utilized to
help establish possible common evolutionary ancestry
for different proteins. For example, the alpha, beta,
delta, and gamma chains of hemoglobin have long
identical sequences (Fitch and Margoliash, 1967). This
argues strongly that the four protein chains were de-
rived from some single gene, perhaps by its endorepli-
cation to form the sort of redundant DNA of which
Britten has spoken, and then the further evolution of
the four genes.

Clearly, the extension of the concept of homology
to the molecular level promises to be exceptionally
rewarding.

Analogy. Analogy is commonly defined as similarity
of function, and is opposed to the evolutionary defini-
tion of homology in terms of common ancestry. It is
often unclear whether analogy is meant to be restricted
to nonhomologous structures. The source of this un-
clarity rests, in part, upon uncertainty whether the
evolutionary, or a phenotypic notion of homology
should be utilized. Granted the evolutionary sense of
homology, it becomes possible to distinguish simi-
larities between organisms which are not due to ho-
mology; thus, a bird's wing and a butterfly's wing are
not homologous despite phenotypic similarity. If we
utilize only a phenotypic concept of homology, it is
unclear how we are to distinguish similarities between
organisms which are homologous from similarities
which are not homologous but analogous. A virtue of
the evolutionary concept of homology, therefore, is
that it allows us to discuss ways in which distinct
(nonhomologous) phylogenetic lines have become
phenotypically similar. The notion of analogy facili-
tates descriptions of phylogenetic convergence.

The notion of analogy can be extended to the mo-
lecular level. One can consider different molecular
structures performing the same function, for example,
different oxygen carrying pigments, or structurally
different enzymes capable of catalyzing the same reac-
tion. The matter is of great importance, for it would
be helpful to have some estimate of the number of
diverse ways in which any chemical (catalytic) job
might be accomplished in order to gain insight into
the difficulty which evolution faced in finding at least
one workable mechanism, or in evolving new ones.
Despite its importance, little work has been done in
this potentially interesting area.

In summary, homology and analogy are working
tools with which the biologist attempts to classify
organisms into hierarchically nested taxa, formulate
phylogenetic hypotheses, discuss evolutionary forces,
describe ontogenetic similarities, and, in short, carry
on his science.

BIBLIOGRAPHY

R. Acher, “Évolution des Structures des hormones Neuro-
hypophysaires,” La Spécificité zoologique des Hormones
hypophysaires et leurs activités, Éditions du Centre National
de la Recherche Scientifique, No. 177, (Paris, 1969).
M. Beckner, The Biological Way of Thought (Berkeley, 1968).
 E . T. Bolton, “The Evolution of Polynucleotide Sequences
in DNA,” Mendel Centenary: Genetics, Development, and
Evolution, ed. R. M. Nardone (Washington, D.C., 1968). R. J.
Britten and D.  E . Kohne, “Repeated Sequences in DNA,”
Science, 161 (1968), 529. D. H. Colless, “An Examination
of Certain Concepts in Phenetic Taxonomy,” Systematic
Zoology, 16 (1967), 7. C. Darwin, On the Origin of Species
by Natural Selection (London, 1859; many reprints). G. R.
De Beer, Vertebrate Zoology (London, 1928); idem, Embryos
and Ancestors, 3rd ed. (Oxford and New York, 1958). W. M.
Fitch and F. Margoliash, “Construction of Phylogenetic
Trees,” Science, 155 (1967), 279. K. Gegenbaur, Grundzuge
der Vergleichenden Anatomie, 2nd ed. (Leipzig, 1870).
R. Goldschmidt, Theoretical Genetics (Berkeley, 1955).
N. Jardine, “The Concept of Homology in Biology,” British
Journal for the Philosophy of Science, 18 (1967), 125.  E . R.
Lankester, “On the Use of the Term Homology in Modern
Zoology,” Annals and Magazine of Natural History, 6, series
4 (1870), 34-43. R. Owen, Lectures on the Comparative
Anatomy and Physiology of the Invertebrate Animals (Lon-
don, 1843); idem, “Report on the Archtype and Homologies
of the Vertebrate Skeleton,” Report of the British Association
for the Advancement of Science, 16 (1847), 169. J. M. Rendel,
Canalization and Gene Control (London, 1967).  E . G.
St. Hilaire, Philosophie Anatomique (Paris, 1818). G. G.
Simpson, Principles of Animal Taxonomy (New York and
London, 1961). R. R. Sokal and P. H. A. Sneath, Principles
of Numerical Taxonomy (San Francisco, 1963). R. R. Sokal
and J. H. Camin, “The Two Taxonomies: Areas of Agree-
ment and Conflict,” Systematic Zoology, 14 (1965), 176.
 E . B. Wilson, “The Embryological Criterion of Homology,”
Biological Lectures, 1894, Marine Biological Laboratory
(Woods Hole and Boston, 1895).  E . O. Wilson, “A Consist-
ency Test for Phylogenies Based on Contemporaneous Spe-
cies,” Systematic Zoology, 14 (1965), 214. L. Wittgenstein,
Philosophical Investigations (Oxford, 1958).

STUART A. KAUFFMAN

[See also Analogy in Early Greek Thought; Biological Con-
ceptions in Antiquity; Evolutionism; Game Theory; Genetic
Continuity; Recapitulation.]