C OMMENTARY

2002 Nature Publishing Group http://www.nature.com/natureimmunology

Historical insight: The clonal selection theory of antibody formation has recently been subjected to challenge from many quarters. A review of its history and that of scientific theories in general points to the
importance of distinguishing between the central hypotheses of a theory and its subsidiary implications.

The Clonal Selection Theory:

what it really is and why modern
challenges are misplaced
Arthur M. Silverstein
Institute of the History of Medicine, Johns Hopkins School of Medicine, 1900 East Monument St, Baltimore, MD 21205, USA. (arts@jhmi.edu)

Like every theoretical statement ... the 1957 theory was made
in terms of contemporary knowledge ... [and is] incomplete ...
[and] expressed in terms that have now become meaningless.
F. M. Burnet (1967)1
The Clonal Selection Theory
(CST)2 of Macfarlane Burnet
and David Talmage seems to
be under fire currently from
several directions. Irun
Cohen3, in considering the
role of autoimmunity in the
economy of the body, suggested that, Progress in
immunology appears to have
rendered the clonal selection
paradigm incomplete, if not
obsolete; true it accounts for
the importance of clonal
activation, but it fails to
encompass, require, or
explain most of the subjects
being studies by immunologists today .... Polly
Sir Frank Macfarlane Burnet. (Courtesy of The
Matzinger4 has gone even
National Library of Medicine.)
further: based upon her view
that the danger theory has
negated Burnets classical view of self-nonself discrimination, she
extends this to suggest that the entire clonal selection paradigm that
has ruled immunology for some 35 years has been overthrown.
It is my contention that the difference between the central hypotheses of clonal selection (which comprise the core theory) and the subsidiary hypotheses (which address the various questions raised by the
theory) have been confused. I shall point out that ideas about tolerance
and self-nonself discrimination are not central to CST, but only stem
from its implications. Thus, there are different levels of hypothesis;
negation of a secondary hypothesis need not threaten the core theory5.
Even philosophers of science have occasionally blurred the important distinction between the core hypotheses of CST and the ancillary
hypotheses that may stem from it. In his book The Immune Self:
Theory or Metaphor?6, A. Tauber calls Burnets view of tolerance ...
a cornerstone of his later immune theory [CST], and throughout
appears to accept the various challenges to Burnets ideas on tolerance

www.nature.com/natureimmunology

september 2002

and self-nonself as challenges to the central meaning of CST. Again,

in their book The Generation of Diversity7, S. Podolsky and A. Tauber
discuss the several challenges to Burnets idea of self-nonself and
conclude that, Specifically, we must ponder whether CST, as constructed by Burnet, Talmage, and Lederberg [sic!8] ... is now being
seriously challenged. Kenneth Schaffner, in his elegant discussion of
the philosophical bases of CST, Discovery and Explanation in
Biology and Medicine, formally defines three levels of hypothesis in
CST and actually assigns Burnets tolerance hypothesis to a secondary level. But even he sometimes seems to suggest that tolerance
experiments may serve as tests of CST9.

From instruction to selection

Between 1930 and the early 1960s, the accepted explanation for the
large repertoire of antibody specificities was that antigen somehow
acts as a template to transfer information to the globulin-producing
mechanism. This was termed an instruction theory and was
advanced in different forms1012. At a time when immunology was preoccupied with chemical approaches13, and when little was known
about how proteins are formed, these instruction theories seemed
plausible. No matter that they could not explain such observations as
the persistence of antibody formation, the accelerated and enhanced
secondary response or what would later be termed affinity maturation.
Then in 1955, in the context of increasing interest in tissue transplantation, immunodeficiency diseases, autoimmunity and tolerance,
Niels Jerne suggested that the information for all antibody specificities is inherent in the host and expressed as natural antibodies14.
Now the sole function of antigen was to select the matching antibody
and transport it to the appropriate cells where, somehow, they would
be stimulated to produce more of the same. Jernes idea attracted little
support at the time; its surviving claim-to-fame lies in the fact that
David Talmage and Macfarlane Burnet independently saw in its selectionist approach the seed of an interesting idea.

The background to clonal selection

The first published response to Jernes idea came from Talmage. He
pointed out that Jernes theory was similar to the side-chain theory of
Paul Ehrlich15, advanced some 60 years earlier, but that Jernes was
improbable in requiring that a protein somehow direct its own replication. Talmage suggested16, rather, that one of the multiplying units of
the antibody response is the cell itself ... only those cells are selected
for multiplication whose synthesized product has affinity for the antigen injected.

volume 3 no 9

nature immunology

793

gen with preformed antibody, and of clonality (hypotheses iii and iv),
Talmage sent a draft of this paper to Burnet, who had already been
involving the cellular dynamics of proliferation and differentiation to
thinking along similar lines. Burnet had been associated with two earyield clones of cells and clones of their product. (Although T cells
lier theories of antibody formation, both expressed in instructionist
were not at the time even on the horizon, we may note in passing how
terms. The first theory17, in 1941, suggested that antigen induces the
reasonably well these hypotheses hold for T cells. Even many of the
formation of adaptive enzymes that would form a protein with a
subsidiary questions will be the same for both systems.)
unique specificity. Even this early, Burnet introduced the idea of the
Each of the core hypotheses raises questions that must ultimately
role of cellular dynamics in the immune response: the knowledge
be answered; speculation about each will involve the formulation of
induced by antigen would persist in all daughter cells that might
subsidiary hypotheses to be tested. Here follow some of the more
result from proliferation. As he put it, ... antibody production is a
obvious questions relating to the core hypotheses (and some contemfunction not only of the cells originally stimulated, but of their
porary answers).
descendants [my italics]. Here was the first hint that cell dynamics
Question (i)a: If a Landsteiner-size repertoire24 arises spontamight be important for antibody formation.
Burnets second theory18, in 1949, proposed that antigen instructs
neously, what is the mechanism for its generation? Subhypothesis
(i)a-1: Burnet suggested25, ... in some stage in early gestation a
the formation of an (RNA?) indirect template against which specificity might be imposed during protein formation. He emphasized again
genetic process for which there is no available precedent, that is, a
the role of cell dynamics in the
rapid somatic mutation. This was
process of antibody formation, but
taken up and expanded elegantly by
now he dealt with a further biologiJoshua Lederberg26, who spoke of
cal complication implied by the thethe rapid somatic mutation of an
d
e
a
f
b
c
ory. If any antigen could stimulate
immunoglobulin gene. This would
an antibody response, why not those
come to be known as the paucigene
Ag.C
native to the host? Peter Medawars
model27. Subhypothesis (i)a-2 (a
studies of tissue transplantation had
somewhat later arrival): Why not a
shown that allografts are rejected,
gene available de novo for each
but autografts are not19. Thus, some
specificity? This would come to be
c
known as the multigene (or
mechanism must exist to distinguish
c
germline) model28. Talmage would
self from nonself. But Ray
f
a
d
e
b
c
Owens demonstration that nonidenanticipate this side of the future
c
c
p
tical cattle twins, connected in
debate in his suggestion that the
c
c
utero, become red cell chimeras
total repertoire is limited, as any
p
c
p
p
unable to respond to one-anothers
Landsteinerian specificity may be
AB .C
antigens20 proved that self-recognidetermined by a unique combination of selected representatives from
tion must be learned rather than
a relatively modest repertoire of
genetically programmed. With the
antibodies29. (In the end, both
suggestion that such a learning
process is important and that it
germliners and somaticists would
30
occurs during fetal development,
Figure 1. Burnets illustration of the CST. Redrawn from F. M. Burnets be proved partly right .)
Burnet foreshadowed his future sinQuestion
(i)b:
If
a
repertoire is
The Clonal Selection Theory of Acquired Immunity, p. 59, Figure 7 (1959), with permission from Cambridge University Press.
gle-minded preoccupation with the
generated randomly and somatiquestion of self and nonself.
cally, why are destructive autoantibodies not formed against native antigens to engender immediate
autoimmune disease? Subhypothesis(i)b: Deserting his earlier
The clonal selection theory
self-marker explanation, Burnet assumed a special susceptibility
Burnet published his theory in, as he later put it with unaccustomed
of immature cells in utero, such that any antigen then present
modesty, ... an exceedingly obscure journal. If the concept proved
would abort that clonal precursor. Lederberg would extend this
to be important, he would have priority; if it were wrong, then ...
idea of a susceptible stage to cover the life of the individual31, as
very few people in England or America would see it21. It had developed from ... what might be called a clonal point of view. Its core
there is no reason why somatic mutation to expand the repertoire
hypotheses may be put succinctly22.
should be restricted to fetal life. (Note that nowhere in his initial
formulation does Burnet mention the terms self or self-nonself
(i) The entire immunological repertoire develops spontaneously in
discrimination.)
the host (that is, there is no information furnished by antigen).
Question (ii): Is there more than one type of receptor on a single
(ii) Each [antibody] pattern is the specific product of a cell and that
cell? Subhypothesis (ii): Burnets hypothesis of clonal deletion
product is presented on the cell surface (as an Ehrlich-type receptor).
(tolerance induction) implied the potential loss of desired specificities
(iii) Antigen reacts with any cells carrying its specific receptor to
if a cell produces many different receptors. Alternatively, the cell
signal cell proliferation and differentiation.
might produce undesired (autoimmune) antibodies when activated,
(iv) Some of these daughter cells differentiate (become plasmacywhat Burnet would term forbidden clones. Burnet thus suggested32
toid) to form clones of antibodies, whereas others survive as clones of
[undifferentiated] memory cells.
that a cell could have reactive sites corresponding to only ... one (or
This, then, is the CST. It was illustrated in its simplicity by Burnet
possibly a small number of) potential antigenic determinants. (Given
in his 1959 elaboration of the theory23, as in Fig. 1. It is a theory of
the above, and the diploid genome, the question of one cell-one antibody would engage the field for a time, see below.)
selection (hypotheses iiii), involving the selective interaction of antiK. R.

2002 Nature Publishing Group http://www.nature.com/natureimmunology

C OMMENTARY

794

nature immunology

volume 3 no 9

september 2002

www.nature.com/natureimmunology

 2002 Nature Publishing Group http://www.nature.com/natureimmunology

C OMMENTARY

Question (iii)a: If somatic mutation persists after clonal expansion,

how can clonal specificity be maintained? Subhypothesis (iii)a:
Burnet does not address this question, but Lederberg suggests33, in his
Proposition A8, that the expanding clone somehow becomes genetically stable.
Question (iii)b: How can interaction of antigen with a surface antibody receptor induce cell proliferation and differentiation?
Subhypothesis (iii)b: It was far too early for Burnet even to ask this
question; it would be decades before the complicated mechanisms of
signal transduction could be unraveled.
Question (iv): What determines and regulates which clonal daughters differentiate to form antibody and which survive as memory
cells? Subhypothesis (iv)a: Burnet does not raise this question. Again,
it is too early to envision cytokine effects, costimulatory molecules,
feedback controls and so on.
Any valid theory of antibody formation (and that is what CST is)
must satisfactorily explain these and other questions. However, the
central theory need not fall just because its promulgator was wrong in
proposing a mechanism to answer one of its subsidiary questions.
Some of these questions address the selection component of CST,
whereas others deal with the clonal component; the former might
survive the disproof of the latter. It is curious that CST has been challenged based upon Burnets error in explaining self-nonself discrimination, but no one has suggested that CST might be challenged because
Burnet was wrong about the mechanism for the generation of diversity,
although these are hierarchically equal hypotheses. Yet self-nonself is
chosen among all the other possibilities, for reasons that we will
explore below.

of T cell receptors (TCRs), Zinkernagel and Doherty42 found that the

TCR reacts with a polypeptide attached to a native major histocompatibility complex molecule. Here was recognition in the context of self,
reinforcing the idea of the sharp divide between self and other.
The second reason for the prevailing interest in self-nonself discrimination is perhaps more important: for many, the phrase selfnonself discrimination has come to epitomize one of the major
unsolved problems facing the discipline today. Most of the other subsidiary questions raised by CST have been clarified fully or in great
measure: the mechanism for the generation of diversity, the nature
and role of T and B cell subsets and their markers, immunoglobulin
class switching, the mechanism of allelic exclusion, the mechanisms
of signal transduction and the nature and role of cytokines and other
pharmacological participants. Still to be defined clearly, however, are
the complex regulatory mechanisms that control the events that follow
the interaction of an antigenic determinant with its T or B cell receptor, those that determine whether the response will be positive or negative, activation or tolerance.
Given this wide-open theoretical terrain, it is no wonder that
debate continues on such questions as a big bang versus the continuous generation of diversity, the relative roles of central versus
peripheral mechanisms of tolerance, the number and type of signals
required for one or the other response43, whether autoimmunity is
dangerous or beneficial (Cohen3) and whether the immune apparatus
evolved to recognize infectious pathogens (Cohn44 and Janeway45),
danger (Matzinger46) or, following Jerne47, self (Coutinho48
and Cohen3).

In the final analysis

Burnets preoccupation with self and tolerance
As early as 1949, Burnet felt the need to explain the acquisition of
immunological tolerance; indeed, his prediction of it as an intrauterine mechanismand Medawars experimental proof of it34won the
1960 Nobel prize for both. Burnet spoke and wrote about it repeatedly, including such books as Self and Notself and The Integrity of the
Body35. With Burnets help, its borders quickly expanded from the
simple explanation of tolerance mechanisms to the evolutionary and
even philosophical implications of self-nonself recognition. Indeed,
immunology has more than once been called the science of self-nonself discrimination36.
Burnet became so involved with the question of self and with his
explanation of the mechanism of tolerance that even he began to view
it as an integral part and even a test of CST, rather than as merely a
subsidiary question to be approached by trial and error. In discussing
the foundations of CST, Burnet admitted that if immunologists are
correct in doubting that tolerance is wholly a matter of the absence
of the immunocyte ... an extensive reorientation [of CST] will become
necessary37. No wonder that others might feel the same! (We might
note also that Burnet came close to giving up on CST in 196238, when
reports came in of two and even four different antibody specificities
produced by a single cell39; when Szenberg et al. found too many
pocks on the chorioallantoic membrane of chick embryos injected
with small numbers of lymphocytes (that is, the proportion of cells
specific for MHC alloantigens appeared to be much too high)40; and
when Trentin and Fahlberg found that a single clone of cells used to
reconstitute a lethally irradiated mouse seemed able to form antibodies of different specificities41.)
There are two further reasons why modern immunologists might
concentrate so much on the immunological self and hold meetings
and symposia on the subject. After the discovery of T cell functions and
www.nature.com/natureimmunology

september 2002

I have attempted to point out the difference between the core

hypotheses that constitute the CST and the ancillary hypotheses
advanced to address the many implications of that theory. It seems
safe to conclude that the two major tenets that comprise the theory
remain unchallenged: antigen selects from among spontaneously
formed antibodies to stimulate certain cells, and this results in the
clonal proliferation and differentiation of these cells. The fact that
Burnet was (at least partially) mistaken in his subsidiary hypothesis
about the mechanism of tolerance induction (clonal deletion in utero)
does not influence the validity of the central theory. On might as well
suggest that Darwins theory of evolution was overthrown by the
demonstration that its author was wrong about one of its important
but subsidiary mechanisms, how variation is inherited: Darwin suggested soft inheritance (the inheritance of acquired characteristics).
We might point out also, to return to Irun Cohens criticism, that like
Burnets CST, Darwins theory also does not encompass, require or
explain most of what evolutionists study today! Such is the nature
of scientific progress.
Even Lederberg recognized early the precise nature and limits of
Burnets theory. He presented nine propositions (hypotheses), of
which four refer to genetics, one to tolerance and three to antibody
formation and memory cells. As he says49, Of the nine propositions
given here, only number 5 is central to the elective theory [my italics]. This is the one that supposes the spontaneous production by a
cell of antibody corresponding to its own genotype. Indeed,
Lederberg even suggested that his elaboration of Burnets explanation
for tolerance (Proposition 6) is not vital to CST and is equally
applicable to instructive theories.
The current preoccupation with mechanisms of tolerance and the
question of self-nonself discrimination appears to have distorted many
views of the nature of CST. It may therefore be appropriate to point

volume 3 no 9

nature immunology

795

 2002 Nature Publishing Group http://www.nature.com/natureimmunology

C OMMENTARY

out that views on this subject cover the spectrum from true believers
to agnostics. A recent extended discussion of the topic revealed that at
least four groups think that self-nonself discrimination is not central
to the problem of immunoregulation and tolerance50.
In the end, however, we must not lose sight of the fact that the CST
is only a theory of how antibodies are formed, not a theory of why
they are formed.

16.
17.
18.
19.

Acknowledgments
I thank K. Schaffner for his helpful discussion of the structure of scientific theories and
N. Rose for useful suggestions.

24.

1. Burnet, F. M. Cold Spring Harbor Symp. 32, 18 (1967), see p. 1.

2. Burnet, F. M. Aust. J. Sci. 20, 6769 (1957);Talmage, D.W. Annu. Rev. Med. 8, 239257 (1957).
3. Cohen, I. R. Immunol.Today 13, 441444 & 490494 (1992). See also Cohens Tending Adams Garden
(Academic Press, San Diego, 2000).
4. The concept first appeared in Matzinger, P. Annu. Rev. Immunol. 12, 9911045 (1994), but CST was
not yet threatened.The claims against CST were exposed most prominently in connection with the
publicity attendant on the publication of three papers from Matzingers and two other labs: Ridge, A.
P., Fuchs, E. J. & Matzinger, P. Science 271, 17231726 (1996); Sarzotti, M., Robbins, D. S. & Hoffman, P.
M. Science 271, 17261728 (1996); and Forsthuber,T., Hualin, H. C. & Lehmann, P.V. Science 271,
17281730 (1996). See also the commentaries by Pennisi, E. Science 271, 16651667 (1996),
Johnson, G. The New York Times p. C1 (March 26 1996) and Dreifus, C. The New York Times p. F4 (June
16 1998).
5. Birgitta Stockinger has also pointed out the exaggeration of these claims for the overthrow of CST
(Immunol.Today 17, 241 (1996)).
6. Tauber, A. I. The Immune Self:Theory or metaphor? p. 93 (Cambridge University Press, New York,
1994).
7. Podolsky, S. H. & Tauber, A. I. The Generation of Diversity: Clonal selection theory and the rise of modern
immunology p. 369 (Harvard University Press, Cambridge, 1997).
8. Lederbergs name has been closely associated with CST because of his paper Genes and antibodies (Lederberg, J. Science 129, 16491653 (1959)). In this, however, he gave genetic substance to
Burnets ideas, but added little to the core hypotheses of clonal selection.
9. Schaffner, K. F. Discovery and Explanation in Biology and Medicine (University of Chicago Press, Chicago,
1993). See also Schaffners discussion of CST in Theor. Med. 13, 175216 (1992).
10. Breinl, F. & Haurowitz, F. Z. Physiol. Chem. 192, 4557 (1930).They pictured the antigen as determining specificity by controlling the order of addition of the amino acids to the polypeptide chain of the
nascent protein.
11. Pauling, L. J. Am. Chem. Soc. 62, 26432657 (1940). Pauling argued that specificity was based upon an
antigen-directed folding of the nascent polypeptide chain to achieve a uniquely molded combining
site.
12. Burnet advanced two instructionist theories. In the context of the times, the firstThe Production of
Antibodies (Macmillan, Melbourne, 1941)involved adaptive enzymes, and the secondBurnet, F. M.
& Fenner, F. The Production of Antibodies edn. 2. (Macmillan, New York, 1949)postulated a genomic
indirect template.These are outlined briefly in the text.
13. For a discussion of the dominant phases of the discipline, see Silverstein, A. M. Cell. Immunol. 132,
515531 (1991).
14. Jerne, N. K. Proc. Natl. Acad. Sci. USA 41, 849857 (1955).
15. Ehrlich, P. Klin. Jahrb. 6, 299326 (1897); also published in English in Collected Papers of Paul Ehrlich
edn. 2, pp. 107125 (Pergamon, London, 1958). Ehrlichs theory is explored at length in Silverstein, A.
M. Paul Ehrlichs Receptor Immunology:The magnificent obsession (Academic Press, New York, 2002).

796

nature immunology

volume 3 no 9

20.
21.
22.

23.

25.
26.
27.
28.

29.
30.
31.
32.
33.
34.
35.

36.
37.
38.
39.

40.
41.

42.
43.
44.
45.
46.
47.
48.
49.
50.

Talmage, D.W. Annu. Rev. Med. 8, 239257 (1957), see p. 247.

Burnet, F. M. The Production of Antibodies (Macmillan, Melbourne, 1941).
Burnet, F. M. & Fenner, F. The Production of Antibodies edn. 2 (Macmillan, New York, 1949).
Medawar, P. B. J. Anat. 78, 176199 (1944); Medawar, P. B. J. Anat. 79, 157176 (1945); also Medawar, P.
B. Brit. J. Exp. Pathol. 27, 914 & 1524 (1946).
Owen, R. D. Science 102, 400401 (1945).
Burnet, F. M. Changing Patterns: An atypical autobiography p. 206 (Heinemann, Melbourne, 1968).
These are not exactly the same as the five slightly modernized and simplified principles given in
Burnets 1968 book Changing Patterns: An atypical autobiography (p. 213), which now included the one
cellone antibody requirement.
Burnet, F. M. The Clonal Selection Theory of Acquired Immunity p. 59 (Cambridge University Press,
Cambridge, 1959).
A Landsteiner-size repertoire refers to the ability of the host to make an antibody to almost any
chemical structure that may be attached as a hapten to a carrier protein, that is, an extremely large
one.
Burnet, F. M. Aust. J. Sci. 20, 6769 (1957), see p. 68.
Lederberg, J. Science 129, 16491653 (1959).
See, for example, M. Cohns discussion in Progr. Immunol. 2, 261284 (1974).
See, for example, Dreyer,W. J. & Bennett, J. C. Proc. Natl. Acad. Sci. USA 54, 864869 (1965); Hood, L.
& Talmage, D.W. Science 168, 325334 (1970). See also Cunningham, A. J. (ed.) The Generation of
Antibody Diversity (Academic Press, New York, 1976).
Talmage, D.W. Science 129, 16431648 (1959).
See Kindt,T. J. & Capra, J. D. The Antibody Enigma (Plenum, New York, 1984).
Lederberg, J. Science 129, 16491653 (1959), see p. 1651.
Burnet, F. M. Aust. J. Sci. 20, 6769 (1957), see p. 54.
Lederberg, J. Science 129, 16491653 (1959), see p. 1652.
Billingham, R. E., Brent, L. & Medawar, P. B. Nature 172, 603606 (1953).
Burnet, F. M. Science 133, 307311 (1961); Burnet, F. M. The Integrity of the Body, (Harvard University
Press, Cambridge, 1962); Burnet, F. M. Self and Not-Self (Cambridge University Press, Cambridge,
1969).
Wilson D. The Science of Self: A report of the new immunology (Longman, Essex, 1971); Klein, J.
Immunology:The science of self-nonself discrimination (Wiley, New York, 1982).
Burnet, F. M. Self and Not-Self p. 30 (Cambridge University Press, Cambridge, 1969).
Burnet, F. M. in Conceptual Advances in Immunology and Oncology pp. 721 (Hoeber-Harper, New York,
1963). See also Szenberg, A. et al. Brit. J. Exp. Pathol. 43, 129136 (1962).
See Nossal, G. J.V. & Mkel, O. Annu. Rev. Microbiol. 16, 5374 (1962). See also Melvin Cohns retrospective review of this debate in Annu. Rev. Immunol. 12, 162 (1994) p. 16ff and Kindt,T. J. & Capra, J.
D. The Antibody Enigma (Plenum, New York, 1984).
Szenberg, A. et al. Brit. J. Exp. Pathol. 43, 129136 (1962).
Trentin, J. & Fahlberg,W. J. in Conceptual Advances in Immunology and Oncology pp. 6674 (HoeberHarper, New York, 1963). Burnet would say (p. 72), This blows out the original clonal selection theory. Ive said before that I dont believe the original clonal selection theory ....
Zinkernagel, R. M. & Doherty, P. C. Adv. Immunol. 27, 1177 (1979).
Bretscher, P. M. & Cohn, M. Science 169, 10421049 (1970); Langman, R. E. & Cohn, M. Scand. J. Immunol.
44, 544548 (1996). See also Langman, R. The Immune System (Academic Press, San Diego, 1989).
Cohn, M., Langman, R. & Geckeler,W. Prog. Immunol. 4, 153201 (1980).
Janeway, C. A. Immunol.Today 13, 1116 (1992); Janeway Jr., C. A., Goodnow, C. C. & Medzhitov, R.
Curr. Biol. 6, 519522 (1996).
Matzinger, P. Semin. Immunol. 10, 399415 (1998) and 4.
Jerne, N. K. Ann. Inst. Pasteur Immunologie C 125, 373389 (1974).
Coutinho, A. et al. Immunol. Rev. 79, 151168 (1984); Coutinho, A., Kazatchkine, M. D. & Avrameas, S.
Curr. Opin. Immunol. 7, 812818 (1995).
Lederberg, J., Science 129, 16491653 (1959), see p. 1649.
These positions by Silverstein, A. M. & Rose, N. R, Grossman, Z. & Paul,W. E., Coutinho, A. &
Cohen, I. R. are elaborated in Semin. Immunol. 12, 159344 (2000). See also Silverstein, M. & Rose, N.
R. Immunol. Rev. 159, 197206 (1997).

september 2002

www.nature.com/natureimmunology