Eleonora Cresto – SHPS C (2008)
1
In search of the best explanation about the nature of the gene:
Avery on pneumococcal transformation
Eleonora Cresto
CONICET (Consejo Nacional de Investigaciones Científicas y Técnicas)
Buenos Aires, Argentina
eleonora.cresto@gmail.com
Published in Studies in History and Philosophy of Biological and Biomedical Sciences,
39 (1), pp. 65-79, 2008.
Available online at: http://dx.doi.org/10.1016/j.shpsc.2007.12.012
Abstract
In this paper I present a model of rational belief change, and I show how to use it to obtain a
better insight into the debate about the nature of pneumococcal transformation, genes and
DNA that took place in the forties, as a result of Oswald T. Avery’s work. The model offers a
particular elaboration of the concept of inference to the best explanation, along decision
theoretic lines. Within this framework, I distinguish different senses in which Avery’s team
can be said to have proceeded with caution, thus throwing some light upon a persistent source
of disagreement among researchers in the history of genetics. In addition, I explain why we are
entitled to say that rival parties such as physicist Maclyn McCarty and biochemist Alfred
Mirsky were epistemically rational, in spite of the fact that they reached different conclusions
on the basis of the same evidence.
Keywords
Inference to the best explanation; Cognitive decision theory; Bayesianism; Avery; DNA;
Pneumococcal transformation
Eleonora Cresto – SHPS C (2008)
2
1. Introduction
In this paper I present a model of rational belief change, and I show how to use it to
obtain a better insight into the debate about the nature of pneumococcal
transformation, genes and DNA that took place in the forties, as a result of Oswald T.
Avery’s work. In very few words, the model offers a particular elaboration of the
concept of inference to the best explanation, along decision theoretic lines. I worked
through the details of this approach elsewhere, so here I shall only sketch the main
features of the proposal, and focus on its ability to carry out a successful analysis of
Avery’s case. In particular, I hope to explain why we are entitled to say that rival
parties such as Oswald T. Avery and biochemist Alfred Mirsky were both, for the most
part, epistemically rational, in spite of the fact that they reached different conclusions
on the basis of the same evidence. In addition, I hope to show that the proposed
framework provides the necessary tools to distinguish different senses in which
Avery’s team can be said to have proceeded with caution, which has been a persistent
source of disagreement among researchers in the history of genetics.
The paper is organized as follows. In Section 2 I state the main controversies
and problems that appear in the literature on Avery’s research; I describe the basic
structure of the model in Section 3, and I apply it to Avery’s case in Sections 4-9, with
the aim of solving some of the problems mentioned in Section 2. In Section 4 I present
a brief historical introduction to bacterial transformation; sections 5 and 6 address
Avery’s attitude, while Sections 7-9 provide rational reconstructions of competing
inferences to the best explanation that might have been carried out by different
Eleonora Cresto – SHPS C (2008)
3
members of the scientific community in the forties and early fifties. Finally, in Section
10 I offer some conclusions.
2. Avery’s case: overview and main problems
Avery and his team are remembered for having stated that the substance capable of
transforming pneumococcal types in controlled experiments was DNA, which in turn
paved the way for the conjecture that genes were made of DNA. 1 Their research
crystallized in three successive papers: Avery et al. (1959 [1944]), and McCarty &
Avery (1946a,b), though the second and third are not as widely read and quoted as the
first one.
Compared to other epoch-making cases from the forties and fifties, Avery’s is
far more complex from an epistemological point of view. Let me mention briefly the
main controversies that have arisen in relation to Avery’s case; by the end of the
section I shall point to those that will be specifically addressed in this paper.
(1) In the first place, the 1944 article is written in a very cautious style, and
hence there is room to discuss whether Avery himself had real and living doubts about
some of his findings. Indeed, many authors seem to suggest that Avery’s style
constitutes proof that he was not convinced of what we take now to be the main moral
1
Strictly speaking, the claim that genes are made of DNA may be said to be a category mistake – for
instance, if we argue that “genes” refer to abstract entities. For a nice discussion of this point cf.
Hotchkiss (1966). In any case, the expression can be taken to be a relatively harmless simplification, and
it is indeed a widely used one, so I shall indulge in it without major qualms. Notice that, to some extent,
the idea of gene as an abstract, non-physical entity survives in contemporary biology, insofar as there is
no privileged way to count segments of DNA. See for instance Kitcher (1992), and Maienschein (1992).
For a brief overview of how the concept of gene evolved, see Carlson (1989), pp. 259 ff. A number of
philosophical considerations related to such evolution can be found in Kitcher (1982), Burian (1985),
and Beurton et al. (2000).
Eleonora Cresto – SHPS C (2008)
4
of his research – namely, that genes are made of DNA; by way of illustration, cf.
Fleming (1968, p. 152); Pollock (1970, p. 14); or Toulmin (1972). For the opposite
view, cf. Diamond (1982), Russell (1988), and, to some extent, Amsterdamka (1993, p.
34). Related to this, some historians sought to distinguish between Avery’s prudent
public statements and his private thoughts on the matter (see Olby, 1994, p. 187; or
Judson, 1996, p. 21 ff.), while others tried to convey a more complex state of affairs,
according to which, even though Avery privately engaged in speculations that went
beyond the content of the 1944 paper, he was also persistently worried about the
possibility of being wrong (cf. McCarty, 1985, especially pp. 163 ff; cf. also McCarty,
1994, p. 394).
(2) Secondly, we find discussions concerning the timing of acceptance of the
relevant hypotheses of the case by the scientific community. To begin with, we might
want to focus on the reactions of other members of Avery’s team. Interestingly,
different collaborators claim to have been committed to different degrees of caution
back then – and, correspondingly, they tend to conceive of such attitudes as the most
reasonable ones; cf. paradigmatically Hotchkiss (1966), for a defense of a moderately
skeptical attitude, and McCarty (1985), for a defense of the reasonability of a bolder
approach back in the forties.
The situation is somewhat more complex when we consider the attitude of
scientist not directly related to Avery’s work. In two well known articles, Wyatt (1972)
and Stent (1972) have claimed, respectively, that Avery’s 1944 paper was not
immediately acknowledged by its peers (it conveyed “information” that did not
become “knowledge” until much later) and that its content was not immediately
Eleonora Cresto – SHPS C (2008)
5
profited by other members of the community (it was a “premature discovery”). Since
then, most writers interested in Avery’s case have rejected the two statements (cf. for
example Lederberg, 1972; Dubos, 1976, pp. 157-159; McCarty, 1985, pp. 227 ff.;
Russell, 1988, p. 393; Olby, 1994, pp. 202 ff; Deichmann, 2004 to mention a few).
Even granting that Wyatt and Stent’s suggestions are ultimately unjustified,
there is still room to discuss whether particular researchers accepted the paper’s
implications sooner or later than others. According to the standard view on this topic
(as found, for example, in Dubos (1976, pp. 145, 156); McCarty (1985, Ch. XII); Olby
(1994, pp. 190 ff.); Cohen (1998, p. 8) scientists such as geneticist André Boivin or
biochemist Erwin Chargaff – to mention two paradigmatic examples – were early
enthusiastic supporters of Avery’s research, whereas biochemist Alfred Mirsky was a
persistent skeptic until much later. This view has been at least partially contested by
Mirsky (1972, 1973). On the other hand, in Watson & Berry (2003, p. 39), we find the
more general claim that geneticists as a whole were readier than biochemists to
acknowledge the consequences of Avery’s article; Hotchkiss (1966, p. 183; 1979, p.
325), by contrast, recalls a very different picture, according to which most classical
geneticists were actually very reluctant to greet Avery’s work with positive eyes.
(3) In the third place, we can find discussions concerning how to assess the
overall role of Avery’s findings for the advancement of genetics. Consider, for
example, the controversy about the legitimacy of the use of the term “revolutionary”
(in some sense akin to the one coined by Kuhn (1970 [1962]) to account for Avery’s
inquiry. In Olby (1990, 1994), for instance, we find an enthusiastic affirmative answer
to the question as to whether the term is applicable in this case; Lederberg (2000, p.
Eleonora Cresto – SHPS C (2008)
6
194), also refers to Avery’s paper as constituting a “scientific revolution”; Pollock
(1970) or Judson (1980), (1996) are good examples of negative answers; while
Dawes’s (2004) paper can be mentioned as an attempt to reach some sort of
intermediate ground (he dubs Avery’s discovery “a quiet revolution”).
(4) Consider also the controversy about the exact relationship between Avery’s
research and molecular biology. Many scientists and historians take Avery’s 1944
paper to mark the origins of molecular biology (cf. for example Burnet, 1968, p. 81;
Wyatt, 1972, p. 86). Other writers, however, while acknowledging the merits of
Avery’s work, prefer to emphasize the role of Luria & Delbrück’s (1943) article on
bacterial mutation, or that of Hershey & Chase (1952) (cf. for example Judson, 1980) –
where all four authors belonged to the so-called Phage Group, which aimed to analyze
genetic replication in bacteriophages. 2 Still other voices, such as Olby (1990), contend
that molecular biology has several roots, including the research done by Avery’s team
and the Phage Group, but also including contributions from scientists working on the
three-dimensional structure of molecules (the “Structural School”), or on plant
viruses. 3
(5) Finally, we might want to reflect on whether the main characters of the case
behaved rationally, given the circumstances. It should be mentioned that there is no
explicit historiographical controversy here; still, I think it is important to bring
attention to the problem, and take a stance on it. Among other things, it relates to
2
For an overview on the Phage research program, cf. Olby (1994), chapter 15; for a first-hand account,
cf. Cairns et al. (1966).
3
For some thoughts on the complex relationship between classical genetics and molecular biology, cf.
Kitcher (2003 [1984]).
Eleonora Cresto – SHPS C (2008)
7
traditional concerns about the extent to which turning points in the history of science
favor the occurrence of decisions that cannot be rationally accounted for.
A quick look at the literature shows that, at times, Mirsky’s attitude has been
suspected to be a by-product of a personal confrontation with Avery’s group (see for
instance McCarty, 1985, p. 148). Similarly, several discussions of Avery’s caution
focus on Avery’s personality (as in Russell, 1988, p. 400; in a somewhat different
sense, cf. also Fleming, 1968, p. 152). Strictly speaking, references to biographical
circumstances of the kind mentioned in Fleming, McCarty or Russell’s accounts are
not incompatible with the existence of more comprehensive explanations that seek to
illuminate whether the relevant agents can be credited with rational epistemic
decisions. However, to this day no such comprehensive explanations have been
proposed to address specifically Avery’s case.
In this paper I shall not attempt to settle contemporary discussions on whether Avery’s
work generated a scientific revolution, or on Avery’s exact contribution to the rise of
molecular biology (points (3) and (4) above). Rather, I shall concentrate on some of
the controversies mentioned in points (1) and (2), and I shall also seek to explain why
we are entitled to say that rival parties in the forties (most clearly, McCarty’s and
Mirsky’s) proceeded rationally, for the most part (point (5)).
Concerning (1) and (2), I shall argue that, in many cases, we find a good deal
of confusion as to what the terms of the discussion really are. For instance, it is not
always clear in the literature which concept of “caution” is at stake. Thus, different
authors that coincide in assessing Avery’s attitude as cautious might be trying to
Eleonora Cresto – SHPS C (2008)
8
convey very different ideas: they might be trying to express that he was legitimately
prudent (as in Hotchkiss, 1966), or that his procedure revealed more skepticism than
reasonably needed (as in Toulmin, 1972).
Other confusions are harder to spell out. I shall attempt to clarify them by
offering a decision-theoretic model of inquiry that requires the elucidation of sets of
questions and potential answers to those questions, whose merits agents are required to
assess.
Concerning point (5), I shall rely again on the aforementioned framework in
order to explain why it was reasonable for McCarty and Mirsky to arrive at different
conclusions, even though they shared the same empirical evidence. I shall also seek to
explain very briefly why other well known models of research are not equally able to
deal with Avery’s case.
A few clarifications are in order. First, it is apparent that real agents, as
opposed to their ideal counterparts, are never perfectly rational: typically, they engage
in sub-optimal ways to gather information (by their own lights), and they fail to
acknowledge useful consequences of their prior beliefs, to mention only a couple of
salient problems. However, we can still wonder whether a given epistemic attitude, as
exhibited by a real agent in the history of science, has a rational core, so to speak –
namely, whether there is a way of connecting the particular epistemic behavior of the
agent with her perception of the epistemic gain, or gain in overall understanding,
involved in such behavior.
Second, I should mention at the outset that I shall not provide direct evidence
for the assumption that the epistemic behavior of rival parties in Avery’s case
Eleonora Cresto – SHPS C (2008)
9
exhibited a rational core, in the sense just stated. I do not think that such a proof can be
offered at all – or demanded, for that matter. As I can see it, on reading the relevant
papers and reports, none of the characters involved comes across as evidently
unreasonable. But, of course, this is not the type of conclusion one can arrive at just by
pointing at a particular passage, or page, within an author’s overall production at a
given period of time. This is not to say that there is nothing we can do to support the
assumption. Indeed, we can obtain an indirect defense by providing a detailed
explanation of why, and how, this situation was possible in the first place. This line of
defense can be reinforced if we show (as I hope to do) that, by engaging in this type of
rational reconstruction, we gain a better understanding of Avery’s case – a type of
understanding that cannot be attained by mere reference to the relevant biographical
data.
In the next section I shall proceed with a brief description of the model of
inquiry that I favor; I shall return to the details of Avery’s research in Section 4.
3. A model of inference to the best explanation
Let me suggest the following model of rational inquiry, within a decision theoretic
framework. I shall describe a research process as a series of steps aimed at the
acceptance of the best explanation available for the set of perplexities of a given agent
at a given time. I shall call it a process of inference to the best explanation, or, for
short, an IBE process.
Eleonora Cresto – SHPS C (2008)
10
Very roughly, an IBE process goes as follows. At a given time t, an agent X is
assumed to have a set of full beliefs KX,t. Then, for an IBE process to begin, researcher
X should be able to identify a set QX,t of questions (not necessarily why-questions) that
X seeks to answer at t. (I shall drop the sub-indices when there is no risk of confusion).
A few formal constraints to QX,t suggest themselves. To begin with, questions
typically have presuppositions. Within the context of semantic analyses of questions,
the label “presupposition” is usually meant to refer to those propositions an agent
should take for granted, as a matter of rationality, if the question is to be meaningful
for that agent. Thus, for example, questions such as “Why φ?,” or “How possibly φ?,”
presuppose that φ does occur. A question such as “What is it that causes ψ?,”
presupposes that ψ occurs (or that it may occur – depending on the case) and that there
is something that causes ψ, and so forth. I shall say that a question q is legitimate for
an agent X at t, if and only if (a) all presuppositions of q of which X is aware at t are in
KX,t, and (b) q is not settled for X at t, that is to say, X does not think that some member
of KX,t counts as a suitable answer to q. Notice that if (a) is false, the question is illformulated, as far as X is concerned.
Suppose q is legitimate for X at t. Then X may, but need not, feel the need to
search for an answer to q at t. She might not be ready to devote time and effort to
attempt to settle q at that particular time –say, because more urgent matters demand
her attention at t. In other words, at a given time t an agent X might only be interested
in answering a small subset of her set of legitimate questions, if at all. In addition, even
assuming she feels the urge to answer q at t, it is not certain that she will be able to
Eleonora Cresto – SHPS C (2008)
11
concoct suitable answers to q: X might only be able to generate possible answers to a
yet smaller subset of the set of legitimate questions that she wants to answer at t.
What is in QX,t, then? QX,t contains all legitimate questions (for agent X, at t)
that X wants to answer at t, and which are able to generate suitable possible answers,
as far as X is concerned. Notice that, so defined, two agents might have identical belief
sets, and hence identical sets of legitimate questions, and still hold different question
sets at t.
As we can see, QX,t is partly defined by reference to a set of answers, which I
shall call set EX,t. Each member of EX,t provides an explanatory answer to all questions
of QX,t; I shall also require that members of EX,t had already been put to the test, and
survived the testing. In addition, I shall demand that all members of EX,t be compatible
with the agent’s prior set of full beliefs; they should be pairwise incompatible; and the
agent should believe that exactly one of them is true.4 Next, agents may choose to
accept some of the hypotheses in EX,t, or they may choose to suspend judgment among
some, or all of them. In order to make a decision as to which expansion strategy to
adopt, the model assumes that agents rely on personal assignments of probabilities and
epistemic utilities over the members of EX,t, and that they seek to maximize expected
4
Here I shall not define what an explanatory answer is. As I can see it, a satisfactory theory of
explanation should focus on offering objective steps to determine when an agent is entitled to select a
best explanation (or, eventually, a disjunction of best explanations) out of a previous batch of
explanatory elements. By contrast, the theory should not tell an agent which hypotheses she should find
explanatory in the first place; this amounts to a non-analyzable, ultimate fact. As I see it, explanatoriness
is directly related to the ability the hypothesis has of enhancing the agent’s overall understanding, in the
sense of promoting a psychologically compelling world picture. And, in this sense, the burden of
deciding which statements qualify as explanatory is on particular agents. The resulting position can be
described as a pragmatist conception of explanation, by analogy with the general layout of pragmatist
epistemology. According to so-called Peircean epistemology, for instance, agents should not seek
justification for their prior beliefs, but only for the changes that occur in their epistemic states; similarly,
according to a pragmatist conception of explanation, agents should not seek justification for finding
certain statements explanatory, but for choosing to expand their belief sets with certain explanatory
statements. For contemporary elaborations on what I call here Peircean epistemology, cf. Levi (1997), p.
4; the Introduction of Fuhrman (1997); Bilgrami (2000), pp. 251-257; or Bilgrami (2004), among others.
Eleonora Cresto – SHPS C (2008)
12
epistemic value, in agreement with the recommendations of (some brand of) cognitive
decision theory. 5
I have argued elsewhere that an adequate concept of epistemic utility on
members of E should be a function of a measure of the virtuosity of the basic
hypotheses. 6 I have in mind features such as simplicity, unification power, fertility,
testability, economy, or accuracy. Then agents are assumed to be able to perform a
trade-off among the many virtues that they deem relevant. Contrary to what seems to
be a common presupposition, I believe that the virtues just mentioned do not bear any
clear relationship with truth or probability. They are not truth-tracking, so to speak –
our most virtuous hypotheses can actually bear low personal probability, and can be, as
a matter of fact, false. Rather, each virtue fosters, in its own particular way, the
construction of a satisfying world-view – and, in this sense, each virtue has its own
peculiar way of promoting understanding. As we shall see in further sections, the
reference to virtues will be crucial at the time of explaining the epistemic behavior of
Mirsky’s and Avery’s parties.
In this paper I shall leave open, for the most part, the exact formulation of the
epistemic utility function that I favor (based on the aforementioned concept of
theoretical virtues), as well as the exact way to interpret the probabilities that enter into
5
The best-known approach to cognitive decision theory is found in Levi (1980); cf. also Maher (1993),
or van Fraassen (1989, 2002). It could be contended that, insofar as I allow for suspensions of judgment,
the set of potential answers for agent A at t is not set EA,t, but the set of all Boolean combinations of
members of EA,t. In this paper I shall not discuss this, or similar, technicalities.
6
Cf. Cresto (2006), chapter 3.
Eleonora Cresto – SHPS C (2008)
13
the equation. 7 However, for the sake of simplicity I shall adopt the idea that, when a
hypothesis H is assumed to be false, its epistemic utility is 0, as no comfort can be
obtained from it. Hence EEU(H) = P(H)f(H)+P(H)0 = P(H)f(H); where P(H) is H’s
probability, f(H) stands for H’s epistemic utility, and EEU(H) is H’s expected
epistemic utility. More complex approaches are of course possible (and perhaps
desirable), but they will not make any substantial difference for the purposes of this
paper.
Within this model, the overall explanatory force of members of E is identified
with their expected epistemic utility. Therefore, whenever there is a single best
element in E, it should be interpreted as the best explanation for the set of perplexities
that prompted the research, as far as the agent is concerned.
What should agents do with best explanations? I shall assume that agents can
be credited with specific thresholds, for particular contexts of inquiry; then, the model
recommends that the best element from E (or the disjunction of all best elements, in
case there is more than one) be added to the agent’s set of full beliefs K – but only if its
expected epistemic utility is above the chosen threshold. As we can see, within this
account the agent’s boldness is conceived of as part of the very structure of what we
should understand by rational epistemic decisions. 8
Notice that, according to this framework, agents come to believe best
explanations because agents think that best explanations are worth the risk. Indeed,
7
I believe the best way to go is to adopt some form of temperate personalism and demand that priors be
sensitive to observed frequencies, when available (cf. Shimmony, 1993 [1970], or Levi, 1994). In
addition, I believe we should not require that agents possess a single personal probability function, but a
convex set of such functions (as in Levi, 1980). But nothing fundamental in this paper depends on the
adoption of this or other perspective on probabilities.
8
Within more complex approaches, caution indices can enter into the composition of the epistemic
utility function; cf. paradigmatically Levi (1984), chapter 5.
Eleonora Cresto – SHPS C (2008)
14
agents risk being wrong – they risk coming to believe a false hypothesis – but taking
the risk may be rational if the gain in overall understanding is high enough. As it is
clear, this view departs from a common way of conceiving of best explanations,
according to which explanatory force is the hallmark of truth. By way of contrast,
within the present model, we are entitled to believe best explanations when their
explanatory force makes the risk of being wrong worth taking.
4. Avery et al. on bacterial transformation
Let me go back to Avery’s case. What did people have in mind, in the thirties and
forties, when they talked about “bacterial transformation,” or, more specifically,
“pneumococcal transformation”? Let me begin by recalling that pneumococci had been
found to come in different specific types; in addition, virulent organisms of any type
were known to be surrounded by a capsule – the reason for the virulence being
precisely that white cells of infected hosts were not able to digest the capsule. Virulent
colonies exhibited smooth edges; hence they were commonly referred to as S
(“smooth”) variants. Attenuated pneumococci, by contrast, were known to be nonencapsulated, and were referred to as R (for “rough”) variants, again due to the visual
form of the colonies. In the twenties Avery had found this capsule to be a
polysaccharide, which was involved in the determination of the pneumococcal type (cf.
Heidelberg & Avery, 1923; Avery & Heidelberg, 1923). This result constituted the
first indication that substances other than proteins could express biological specificity,
against the common wisdom of the time.
Eleonora Cresto – SHPS C (2008)
15
Frederick Griffith (1928) reported that mice inoculated with heat-killed type I S
pneumococci together with a live R II strain died, surprisingly, from a type I S
infection; Dawson & Sia (1931) were able to repeat the transformation experiment in a
test tube, and soon after that, Lionel Alloway – a younger colleague of Avery’s at the
Rockefeller Institute – showed how to obtain transformation in the presence of active
extracts of the heat-killed pneumococci, rather than using intact, heat-killed cells
(Alloway, 1932, 1933).
Griffith had interpreted the transformation phenomenon as an indication that
pneumococcal types might change in response to environmental conditions. Against
this, Avery’s earlier studies on the pneumococcal capsule pointed to the fixity of
immunological types – hence Avery was of course intrigued by the chemical nature of
the transforming agent. On the other hand, it was hard to imagine back then that it
would turn out to be DNA. In (1931) Levene and Bass had proposed that the DNA
molecule exhibited a repeating structural unit, represented by the four nucleotides
arranged in the same order. In Chargaff’s words, it was a well-established dogma
before his 1950 paper appeared (cf. Chargaff, 1950, 1979, pp. 350-351). Thus, DNA
was conceived of as devoid of biological specificity, which implied that it could not be
the trasnsforming substance. Proteins, by contrast, appeared to be the natural
candidates to play this role.
The experiment described by Avery, MacLeod and McCarty in 1944 accounted
for the transformation of rough type II pneumococci into smooth, virulent type III. In
very few words, the experiment consisted in placing a culture of RII pneumococci in a
suitable medium, together with a substance (the “transforming principle”) capable of
Eleonora Cresto – SHPS C (2008)
16
inducing type transformation. Once a change of type was effectively seen to occur, the
transforming substance would be analyzed to try to establish what it contained, and
what it did not.
As I have already suggested, our understanding of Avery’s case can benefit
from the application of the model sketched in the previous section. In order to do so,
let me begin by considering two main questions that caught the attention of researchers
on pneumococci at the Rockefeller in the early forties:
(i)
What is the chemical composition of the transforming substance? (For
example: Protein? RNA? DNA?)
(ii)
How should we interpret the activity of the transforming substance?
And, correspondingly, what does the transformation phenomenon really
amount to? (For example: Does it reflect the action of a virus? Of a
gene?)
Notice that, in the context of discussions on bacterial transformation, in order to assert
that genes are made of DNA we would need to accept both the claim that the
transforming principle reveals the action of genes, and the claim that it is DNA. That is
to say, we would need to answer both questions (i) and (ii) in a particular way. But,
insofar as (i) and (ii) express two distinct, independent problems, each of which
requires a particular set of possible solutions, the timing of acceptance of answers to (i)
and (ii) need not coincide. Related to this, considerations about Avery’s putative
caution should better specify whether we are referring to Avery’s attitude to (i), or to
(ii), or to both.
In the next section I shall address Avery’s research from his own point of view.
Eleonora Cresto – SHPS C (2008)
17
5. Avery’s question set
Let QA be Avery’s question set in the thirties, which prompted the research that led to
the 1944 paper and its sequels. We can say that the content of Griffith’s 1928 report
acted as the surprising fact that called for an explanation, in Peircean terms. 9 But what
was it exactly that Avery sought to explain? We read, for instance:
[Our] major interest has centered in attempts to isolate the active principle from crude bacterial
extracts and to identify if possible its chemical nature or at least to characterize it sufficiently to
place in a general group of known chemical substances. (1944, p. 175).
This is no other than question (i) from Section 4. Consider also Avery’s well known
letter to his brother Roy:
[A]fter innumerable transfers and without further addition of the inducing agent, the same
active and specific transforming substance can be recovered far in excess of the amount
originally used to induce the reaction. Sounds like a virus – maybe a gene. But with
mechanisms I am not now concerned – One step at a time – and the first is, what is the
chemical nature of the transforming principle? Someone else can work out the rest.
(Reproduced in McCarty, 1985, p. 159; my emphasis).
Again, here Avery asserts explicitly that his first preoccupation is to find an answer for
(i). Moreover, in this passage he states clearly that at the time he is not concerned with
the kind of question that may be answered by claiming that the transforming principle
is “a virus – maybe a gene.” Hence, question (ii) as stated in the previous section is not
in QA.
This asymmetry between (i) and (ii) will turn out to be consequential at the
time of assessing whether Avery did, or did not, proceed with excessive caution. Let
9
In the sense of Peirce (1931-1958), Vol. 5, p. 189.
Eleonora Cresto – SHPS C (2008)
18
me devote the remaining part of this section to deal with question (ii); I shall postpone
a discussion of Avery’s attitude towards the potential answers to (i) until Section 6.
Was Avery too cautious at the time of interpreting the activity of the
transforming substance? In the light of the above, being cautious in this respect cannot
mean failing to accept a preferred potential answer to (ii) (or, at any rate, a non-trivial
disjunction of preferred potential answers), insofar as (ii) was not a question of QA.
However, someone could complain that, by neglecting (ii), Avery revealed a serious
lack of understanding of the genetic implications of the case. According to this line of
argument, had Avery realized that the transformation process might well have
amounted to a genetic change – and had he understood the enormous consequences
that this fact had for biology – he would not have refused to take question (ii)
seriously, and he would have risked asserting an appropriate answer to it.
At least at first blush, this criticism seems wrongheaded. There are many
sources that document that Avery was perfectly aware of the possibility of a
connection between his experiments on transformation and regular genetic facts. In
addition to the aforementioned letter to his brother Roy, in 1943 he reported:
The genetic interpretation of [the transformation] phenomenon is supported by the fact that
once transformation is induced, thereafter without further addition of the inciting agent both
capsule formation and the gene-like substance are reduplicated in the daughter cells. (Avery &
Horsfall, 1943, pp. 151-152).
And, in 1947,
…those of us actively engaged in the work have for the most part left matters of interpretation
(of the transformation phenomenon) to others and have chosen rather to devote our time and
thought to experimental analysis of the factors involved in the reaction. This is not to say that
we are indifferent and have not among ourselves indulged in speculation and discussion of the
relation of the problem to other similar phenomena in related fields of biology. (Avery, 1947,
p. 127)
Eleonora Cresto – SHPS C (2008)
19
In the light of previous reports, by “similar phenomena in related fields in biology” he
most certainly meant to include genetic phenomena. Incidentally, the quote from 1947
also shows that Avery voluntarily chose to focus exclusively on the attempt to settle (i)
– hence reinforcing my previous suggestion that (ii) was not in QA.
Still, we owe the potential critics some explanation regarding how it could have
been reasonable for Avery to suspect that the transformation process was the action of
genes, and yet refuse to address (ii). We can produce the required explanation by
recalling that, according to the model presented in section 3, a question may be
legitimate for an agent and still not be the type of question that prompts an IBE
process, as far as the agent is concerned. But reluctance to perform a particular IBE
process at a given time does not show that an agent is ignorant of the potential
implications of her current research, or that she is not interested in solving other
questions beyond the ones that prompted the original inquiry. In particular, I have
suggested that no question set Q can be defined unless the agent is able to identify a
suitable set E of pairwise incompatible answers to the members of Q. Indeed, we might
well conjecture that (ii) was not in QA because Avery did not feel comfortable at
building a suitable EA for a question set that contained both (i) and (ii) – the reason for
this being, in turn, that he had not concocted specific experiments to test the various
possible responses to (ii) that might have occurred to him in the early forties: he was
just too busy searching for the chemical composition of the transforming principle. So
even though in 1944 Avery and his co-workers did list the several solutions to (ii) that
were popular in the literature back then (see Avery et al., 1959 [1944], p. 190), they
never gave any indications to the effect that they thought of the aforementioned list of
Eleonora Cresto – SHPS C (2008)
20
option as exhaustive, or as serious candidates for explanation by their own lights.
Avery must have felt that he was not in the position to start an IBE process on such
bases. Rather, he might have hoped to know more about the chemical composition of
the principle (and about the transformation phenomenon as a whole) before embarking
in a new IBE. This is not incompatible, however, with his believing that, were he to
ask question (ii) with the goal of an IBE in mind (and had he conducted prior specific
tests to this effect), a genetic answer of some type would probably have been among
the favorite ones.
A new counterargument suggests itself: shouldn’t the very same research that
Avery and his collaborators were conducting, by its very nature, have shed light on all
the relevant elements needed to perform a suitable IBE regarding question (ii)?
Against this suggestion, I believe Avery’s attitude towards (ii) was sensible, given the
circumstances; to put it differently, he was indeed cautious, but certainly not overly
cautious. The fact that changes induced by DNA were “predictable, type-specific and
heritable” (Avery et al., 1959 [1944], p. 190) strikes our contemporary sensibility as
more than sufficient evidence to talk about genes. But relating the pneumococcal
transforming principle to genetic activity did not come naturally to anyone back then,
as it was not even clear that bacteria had genes. As Hotchkiss put it nicely,
[Back in 1940], genetists, too, could not get past the objection that one whole bacterial cell took
part in making two daughter cells; so they found no sign of the channeling of genetic
determinants through such a concentrated stage as a chromosome. It may have seemed to some
unfair for them to ask for a mating test to demonstrate the genes in bacteria – but without it,
where was the evidence that specific determinants exist which sometimes do, and sometimes do
not, manage to gain access to a particular cell? (Hotchkiss, 1966, p. 183)
Eleonora Cresto – SHPS C (2008)
21
[A classical geneticist] might have asked me to show that our bacteria had compound eyes, or
two sets of wings (Hotchkiss, 1979, p. 325)
Even if the presence of genes in bacteria were not an issue, the way rough
pneumoccocus of type II led to smooth type III was not an obvious example of sexual
cross, so it took some time for biologists and geneticists to understand what was going
on. 10
Let me turn now to question (i).
6. Competing hypotheses
How shall we reconstruct the set of basic explanations EA built as responses to
question (i)? As time went by, different hypotheses were suggested, prompted by the
need to purify the transforming substance by discarding from the killed type III cells
what the transforming principle was probably not. As a result of the process of
proposing specific conjectures and putting them to a test, Avery and his co-workers
ended up with a definite set of options, with specific probabilities attached. 11 The
transforming principle was taken to be:
10
11
H1
Capsular protein (a type-specific antigen)
H2
A polysaccharide
H3
RNA
But more about this in Section 7.
Here I adopt the view that tests very seldom prove the falsity of given hypotheses – the most they do
is make them very unlikely.
Eleonora Cresto – SHPS C (2008)
H4
22
A protein located in the nucleus of the cell (that is, the “protein
version of the central dogma” in the theory of the gene, in Olby’s
terminology) 12
H5
DNA plus protein (that is, nucleoprotein – or Mirsky’s
“chromosin”)
H6
DNA alone
Thus, let EA={H1-H6}. Recall that all members of an agent’s set of basic hypotheses
should be compatible with the agent’s set of full beliefs. I shall leave open the
possibility that Avery might have contracted at some point his belief set KA, so as to
avoid conflict with H1-H6. For example, I shall assume that, had Levene’s
tetranucleotide hypothesis been in KA in the thirties, it would have been removed well
before putting H6 to the test, and hence well before building EA. Mutatis mutandis for
any other statement incompatible with the members of EA.
Interestingly, there is no agreement on when exactly Avery and his co-workers
begun to focus on DNA – not even among the main characters involved. 13 But, in any
case, at some point H6 emerged as the favorite candidate. Did Avery actually accept H6
by the mid-forties? There is no straightforward answer to this question, so let us
examine the evidence with some care.
On the one hand, several retrospective accounts suggest that Avery was still
skeptical by 1943 (that is, by the time the 1944 paper was submitted for publication):
I can remember that as we discussed the situation on the way home on the train, Colin asked
him with a certain amount of impatience: ‘What else do you want, Fess? What more evidence
12
13
Cf. Olby (1994), Ch. 6.
For instance, compare Hotchkiss (1965), p. 5 – and also Olby (1994) p. 185 – with McCarty (1985), p.
232.
Eleonora Cresto – SHPS C (2008)
23
can we get?’ I don’t believe that he replied to this, but one answer that he had was to seek still
more advice. (McCarty, 1985, p. 163) 14
I am afraid that both Colin and I became increasingly impatient with Avery’s caution, even
though we were not unaware of the importance of being sure of our ground. We were just
young enough to become convinced more readily. Avery expressed his doubts repeatedly in his
letter to [his brother] Roy [in May 1943] and they were also obvious on almost a daily basis in
the laboratory. (McCarty, 1994, p. 394) 15
On the other hand, it is not clear that we can take the style of the 1944 paper – which
was undoubtedly Avery’s (cf. for instance McCarty, 1985, p. 165) – to support the
claim that Avery was overly cautious. (Notice that Avery might have indeed been
cautious by 1943, as shown by McCarty’s memoirs, and still be the case that the paper
does not constitute additional proof of this). Surely, the authors never stated that H6
was actually true, but this omission is not self-explanatory. What they did say was that
the transforming principle consisted “largely, if not exclusively,” of DNA, and that the
results “strongly suggested” that DNA possessed specificity (1994, p. 188). In another
widely commented upon passage near the end of the discussion section, we find:
It is, of course, possible that the biological activity of the substance described is not an inherent
property of the nucleic acid but is due to minute amounts of some other substance adsorbed to
it or so intimately associated with it as to escape detection. (Ibid, p. 190)
But soon after that, at the closing paragraph we read, once more:
The evidence presented supports the belief that a nucleic acid of the desoxyribose type is the
fundamental unit of the transforming principle of Pneumococcus Type III. (Ibid, p. 191)
14
Compare with McCarty (1994), p. 394, where he states that “the ‘else’ that [Avery] would have liked
to have was a purified DNase to try on the transforming DNA.”
15
Curiously, both Olby (1994) and Judson (1996) take Avery’s letter to Roy to support the
interpretation that Avery was publicly cautious, but privately confident of H6.
Eleonora Cresto – SHPS C (2008)
24
In other words, the available evidence “supported the belief” that H6 was correct –
although they did not say, literally, that it was correct. This, however, by itself, is
hardly sufficient evidence to conclude that Avery was still unconvinced. The style is
not unusual for a scientific article; we can find well known examples in which similar
writing styles were deemed adequate – by the relevant portion of the scientific
community at the time – to convey particular results loud and clear, and were not
perceived as symptoms of excessive prudery. 16 In the light of this, we might wonder
whether the perception, in the mid-forties, of Avery et al.’s 1944 paper as unusually
cautious was not the result of fellow scientists’ inadvertently imposing well known
facts about Avery’s personality to the paper itself. After all, as many writers have
pointed out in response to Wyatt’s claims, news was transmitted by personal
interaction as much as by publications in scientific journals (cf. for instance Olby,
1994, p. 202).
In short, there are good reasons to say that Avery still had doubts about the
truth of H6 by 1943, even though such reasons do not include considerations about the
style of the 1944 paper. In addition, Avery might have become more confident of H6 as
16
Just to mention a couple of cases, in Lederberg and Tatum’s famous 1946 paper we read that their
experiments “strongly suggest the occurrence of a sexual process” in Escherichia coli (1946, p. 558; my
emphasis) – yet their writing style never prompted any controversies. The style is not peculiar to biology
either. For a very well known example drawn from the history of physics, we might recall J. Chadwick’s
celebrated report in (1932), which is usually understood as an attempt on Chadwick’s part to recount the
observation of a neutron, even though Chadwick explicitly states that it is not easy to choose between
the hypothesis that a neutron has been observed, and a competing hypothesis. [I want to thank Alejandro
Cassini for this example]. In general, I am inclined to think that any paper that puts forward a particular
answer as the favorite candidate to settle a particular question – say, by stating that it is comparatively
more supported by the evidence – can be understood as urging the reader to perform a suitable IBE, and
(in case the reader holds the appropriate priors and epistemic utilities), accept the hypothesis. The 1944
article by Avery, MacLeod and McCarty satisfies the antecedent of this general maxim: the authors took
pains to establish how much the evidence boosted the probability of H6, whereas they raised no doubts
about its epistemic virtuosity. In other words, the paper was clearly structured so as to instill confidence
in the truth of H6 – the prior of H6 being each reader’s responsibility. Here I shall not press this line of
argument any further.
Eleonora Cresto – SHPS C (2008)
25
time went by and he managed to gather more evidence, as he certainly had done by
1946:
The results of the present investigation show that in order to detect proteolytic activity, it is
necessary to use an amount of purified desoxyribonuclease 100,000 times greater than that
required to cause rapid and complete destruction of activity of the transforming substance. This
evidence, in conjunction with the data previously reported on the chemical and physical
properties of the active principle, leaves little doubt that the ability of a pneumococcal extract
to induce transformation depends upon the presence of a highly polymerized and specific form
of desoxyribonucleic acid, and that this constituent is the fundamental unit of the transforming
principle. (McCarty & Avery, 1946a, p. 94)
Let me state briefly what we have obtained so far. It is clear that Avery did not believe
that genes were made of DNA by the mid-forties. Can we accuse him of being unduly
cautious? I have argued that, in order to address this problem, we need to distinguish
between Avery’s willingness (or lack thereof) to accept that the transformation
phenomenon revealed the action of genes (which counted as a possible answer to (ii)),
and his willingness to accept that the transforming principle was DNA (which was
meant to answer (i)). We have seen that, regarding (ii), we cannot accuse him of being
too prudent because of reluctance to endorse an explicit connection between bacterial
transformation and genetics, but, if at all, because of a more fundamental reluctance to
conceive of (ii) as a question that could prompt a suitable IBE process in the early
forties. But, first, from this fact we should not infer that he was ignorant of the
relevance of the problem, or of its potential implications; secondly, I have argued that
the caution Avery exhibited by not adopting question (ii) as an IBE-prompting
question was perfectly justified at the time. On the other hand, by 1943 Avery seemed
not to have fully believed H6 either, even though he was explicitly committed to
Eleonora Cresto – SHPS C (2008)
26
finding an answer to (i); he most probably became convinced around 1946. As we
have seen, the two situations are not comparable: by 1943 we could not have expected
him to accept the idea that the transforming principle was linked to genetic activity,
although we could have hoped him not still to be a skeptic about H6. Are we entitled to
say that he overly cautious with respect to H6, then? To answer this question, we
should assess first whether H6 carried highest expected epistemic utility, from Avery’s
own perspective. I shall postpone a discussion of this point until Section 8, where I
shall also discuss the epistemic behavior of other members of his team.
7. Accepting that DNA is the “transforming substance”
In this section I want to discuss the timing of acceptance of H6 by various members of
the scientific community; I shall provide suitable explanations for their epistemic
behavior in Sections 8 and 9.
We have already seen that Avery might have accepted H6 by 1946 (although, as
a matter of fact, there is no unequivocal first-person account stating so). The situation
is less controversial in the case of McCarty, Boivin or Chargaff, to mention a few clear
examples. McCarty’s commitment to H6, for instance, can easily be revealed by the
content of his lectures right after 1944:
Certainly, there can be little doubt that desoxyribonucleic acid must be present in its intact,
highly polymerized form [for transformation to occur], and when all of the evidence is
considered it appears extremely unlikely that small traces of some other specific substance,
such as protein, could be responsible for the manifestation of transforming activity. (McCarty
1946; quoted in McCarty, 1985, pp. 206)
Eleonora Cresto – SHPS C (2008)
27
and, more emphatically, by the general tone of his memoirs (McCarty, 1985, 1994).
Consider also some of Boivin’s claims from the 1947 Cold Spring Harbor Symposia,
where he presented results of tranformation in Escherichia coli:
In bacteria – and, in all likelihood, in higher organisms as well – each gene has as its specific
constituent not a protein but a particular desoxyribonucleic acid which, at least under certain
conditions (directed mutations of bacteria), is capable of functioning along as the carrier of
hereditary character; therefore, in the last analysis, each gene can be traced back to a
macromolecule of a special desoxyribonucleic acid. (Boivin 1947, p. 13)
Notice that by 1947 Boivin appeared to be not only convinced of H6, but of the fullfledged statement that genes were made of DNA, whereas he mistakenly interpreted
the transformation phenomenon as “directed mutation of bacteria.” A similar attitude
can be attributed to Chargaff, who asserted in many occasions that Avery’s paper
immediately made him change the course of his research to focus on DNA (cf. for
instance Chargaff 1978, pp. 82 ff.).
In short, several scientists became committed to the truth of H6 no later than by
1946-47. In McCarty’s case, moreover, his “impatience” with Avery’s caution prior to
1944 (cf. quote on p. 21) point to the fact that acceptance goes back at least to 1943.
Other researchers, by way of contrast, were clearly wary of H6 during the
forties. Most evidently, Mirsky expressed serious concerns on whether all traces of
protein had actually been removed:
There can be little doubt in the mind of anyone who has prepared nucleic acids that traces of
protein probably remain in even the best preparations… No experiment has yet been done
which permits one to decide whether this much protein [1 or 2 per cent] actually is present in
the purified transforming agent and, if so, whether it is essential for its activity; in other words,
it is not yet known which the transforming agent is – a nucleic acid or a nucleoprotein. To
claim more, would be going beyond the experimental evidence. (Mirsky & Pollister, 1946, pp.
134-135)
Eleonora Cresto – SHPS C (2008)
28
Similar claims can be found in Mirsky (1947, p. 16); Mirsky & Ris (1949, p. 667); or
Mirsky (1951, p. 133) – hence showing that he was still skeptic around the early fifties.
It is worth mentioning here that some of his colleagues considered that his doubts were
malicious (cf. for example Perutz, 1994), and that they prejudiced part of the scientific
community against Avery’s research (cf. McCarty, 1985, p. 218; cf. also Olby, 1994,
pp. 192-193). On the other hand, according to Mirsky’s own recollection of the
situation,
In 1946 and 1951 I accepted the idea that DNA is part of the transforming material, but asked
whether protein is not also necessary. At the time this was an obvious question. It was finally
decided by Hotchkiss’s work and in 1953 I do not mention the possibility of protein still being
there. (Mirsky, 1973).
Interestingly, Mirsky has also stressed that, in his view, it would be definitely unfair to
regard him as one of the chief opponents to Avery’s conclusions (Mirsky, 1972, 1973).
As it turns out, however, what he meant by this is that he had succeeded in
understanding the role of the transforming agent sooner than people more directly
involved with Avery’s project – in particular, sooner than Hotchkiss: by 1951
Hotchkiss was still thinking of the transformation phenomenon as the induction of a
specific mutation, while Mirsky already considered it “to be essentially a
hybridisation.” (Mirsky, 1972; cf. also Mirsky, 1951, p. 133). In other words, Mirsky
can be shown to have been committed early on to the right answer to (ii) – although it
should be clear that this, by itself, does not tell us anything relevant about Mirsky’s
attitude towards H6.
Eleonora Cresto – SHPS C (2008)
29
In the next couple of sections I shall seek to offer a rational reconstruction of different
epistemic attitudes towards H6. In particular, I shall attempt to explain McCarty, Avery
and Mirsky’s behavior, which, as we have seen, amount to cases of early acceptance,
early caution, and late skepticism, respectively.
8. In search of a suitable explanation (I): McCarty and Avery
How shall we explain the epistemic behavior of members of Avery’s team by 1943?
Consider first the probabilities that might have been at stake. It is plausible to say that
Avery, MacLeod and McCarty’s personal probability functions did not differ too much
from one another, insofar as the cumulative effect of the tests reported in the 1944
paper tended to wash priors away. (This would occur even more so as new
experiments were conducted in the late forties and early fifties, but the phenomenon
was still sufficiently clear before their first paper was published.) Let PA refer to such
probability measure. Indeed, by the time EA was built, the probability of all elements of
EA had been suitably updated in response to the results of many successive tests: as
Avery’s research program made progress, the probabilities of H1, H2, H3 and H4 went
down, whereas those of H5 and H6 became larger. Moreover, the probability of H6
augmented more than that of H5, given that H6’s likelihood could well be taken to be
higher than that of H5 (i.e., Pi(ei/H6) > Pi(ei/H5), where “ei” refers to the result of test i,
and “Pi” refers to the personal probability function held by members of Avery’s team
right before updating by ei).
Eleonora Cresto – SHPS C (2008)
30
Let me focus now on the way Avery, MacLeod or McCarty could have
assessed the epistemic virtuosity of the hypotheses at stake. To a large extent, this
depends on which statements were in each agent’s set of full beliefs at the moment of
evaluating virtues; I think it is safe to assume that the relevant portions of Avery,
MacLeod or McCarty’s belief set (for the present context) did not differ significantly
from one another. Consider, first, which elements of EA might have ranked best at
unification power. In principle, many different hypotheses can enhance unification,
albeit in very different senses. For example, stating that the transforming principle is a
polysaccharide may lead to a picture in which the generating substance and that which
is generated (a type-specific capsule in pneumococci) are chemically alike; stating that
it is a protein, on the other hand, matches well with the importance attributed to
proteins in the early forties. As there is not a clear winner, let me turn to the one which
ranks definitely worst. H5 is a clear candidate for this: it refers to two substances rather
than one, it does not take tests at face value (the additional reference to proteins in
addition to DNA does not seem to play any explanatory role, as far as Avery,
MacLeod or McCarty are concerned) and, in general, it presents a more complex
picture of bacterial transformation. 17
Similar considerations indicate that H5 ranked worst at testability and economy,
from Avery’s team’s point of view, insofar as it was uncertain how to test the presence
of substances that could not be detected by state-of-the-art techniques. As for fertility,
17
This analysis, however, might not hold for other researchers: my claim that there is no overall
unification gain in accepting complex hypotheses such as H5 might not be true if the prior set of full
beliefs were somewhat different, as we shall see.
Eleonora Cresto – SHPS C (2008)
31
H6 was, as a matter of fact, the only hypothesis that led Avery, McCarty and (later)
Hotchkiss to design a clear path of future research.
In short, there was at least one virtue at which H6 did better than any other
hypothesis (namely, fertility), and quite a number of other virtues at which its most
important rival (to wit, H5 – Mirsky’s favorite candidate) did worse.
Let me show now how to integrate these elements so as to produce a
comprehensive explanation of particular epistemic attitudes. Consider first McCarty’s
behavior. I have suggested that McCarty’s personal probability for H6 was larger than
that of its rivals. But notice that, by itself, this fact does not constitute sufficient reason
to say that he was justified in believing it. By way of illustration, if all McCarty cared
about was to avoid importing error into his belief set, then the best option for him
would have been to suspend judgment. However, as a matter of fact, researchers also
care about enhancing their understanding of the world – or else no research project
would ever get off the ground. In McCarty’s case, we can provide a rationale for his
epistemic attitude towards H6 by taking simultaneously into account his evaluation of
how probable H6 was, as well as his assessment of H6’s overall virtuosity (to the best
of our knowledge) – in short, by taking into account McCarty’s assessment of H6’s
overall explanatory force. On the basis of such considerations, it seems reasonable to
postulate the existence of an appropriate epistemic utility function fA(Virtue1(Hi),…,
Virtuej(Hi)), such that EEU(Hi) = PA(Hi)fA(Hi) was maximum for H6, and such that
EEU(H6) was above McCarty’s particular caution threshold.
How shall we explain Avery’s epistemic behavior, then? According to the
present analysis, it seems reasonable to say that EEU(H6) was maximum for Avery as
Eleonora Cresto – SHPS C (2008)
32
well – however, we might also conjecture that his acceptance threshold was just
stricter than McCarty’s. In what sense – if at all – can we then assert that he was being
overly cautious? I take it that such a claim should be understood as conveying a
relative sense of caution – relative to the speaker’s own boldness; in other words, I
suggest that, within the present analysis, claims to the effect that Avery was
excessively prudent should be taken to mean that, had the speaker been in Avery’s
shoes – and had the speaker shared Avery’s assessment of H6’s expected epistemic
utility – she would have had a higher acceptance threshold. 18
9. In search of a suitable explanation (II): Mirsky’s case
As we have seen, it is safe to assume that Mirsky was, at the very least, still skeptic
about H6 by 1952. In addition, we can also assume that Mirsky had already rejected
H1-H4 by the late forties; recall that, as early as in 1946 he stated explicitly that the two
serious options were H5 and H6. 19 So let Mirsky’s set of basic explanations be
EM={H5, H6}.
Consider now the IBE process that might have taken place on the basis of EM.
Why did H5 “survive” for Mirsky by the early fifties, and not for Chargaff, Boivin, or
18
The problem of whether there is any “objective,” or “absolute” sense of caution is indeed an
interesting one (can we ever say that some caution indices are just too high to be rationally admissible?)
but well beyond the scope of this paper.
19
“[I]t is not yet known which the transforming agent is – a nucleic acid or a nucleoprotein.” (Mirsky &
Pollister, 1946, p. 135).
Eleonora Cresto – SHPS C (2008)
33
Avery’s team, to mention a few? To address this question, let me start by conjecturing
Mirsky’s personal probability function by that time, or PM.
Notice that right after 1944 we could expect that Mirsky takes H5 to be more
probable than H6. Whatever might have been the evidence that led him to endorse
Levene’s hypothesis in the past (and the lack of specificity of nucleic acids), it could
also count as evidence in favor of H5 and against H6 – even if Levene’s hypothesis
were no longer in Mirsky’s belief set. To this we should add a possible lack of
confidence in Avery’s biochemical skills, which might have led him to question the
significance of Avery’s results.
In the light of the above, we might be tempted to say that acceptance of H6
occurred as soon as new evidence in favor of H6 came in. In this sense, we could
interpret Mirsky’s statements in (1973) (cf. page 26 above) as expressing that, once he
got acquainted with Hotchkiss’s results in (1952), the probability of H6 became
sufficiently high so as to justify acceptance. I believe, however, that this reconstruction
is misleading, for it omits part of the story. Let me explain why.
To begin with, recall that by July of 1948 Hotchkiss had already presented the
crucial findings that would constitute the core of his (1952) paper – namely, that the
ratio of contamination of the transforming principle with protein was less than 0.02%
(Hotchkiss, 1952; cf. Hotchkiss, 1949). Even assuming that Mirsky had not heard
about Hotchkiss’s research until 1952 – and even disregarding Boivin’s 1947 report at
the Cold Spring Harbor Symposium on Escherichia coli (which could not be
confirmed by other laboratories) – many other relevant results were obtained during
those years. Consider, in the first place, McCarty and Avery’s papers in 1946
Eleonora Cresto – SHPS C (2008)
34
(McCarty & Avery, 1946a,b), which showed that minute amounts of purified DNase
inactivated the transforming agent, that a fivefold greater yield of purified
transforming agent could be obtained by inhibiting the action of DNase with citrate,
and that the transforming substance of types II and IV behaved just as that of type III.
We should also count a number of crucial tests on the relation between genes and
DNA, such as Boivin et al. (1948), as well as Mirsky’s own studies with Hans Ris in
(1949, which showed that the amount of DNA in diploids cells doubled that of
haploids cells (Mirsky & Ris, 1949). Let me also add to the list Chargaff’s (1950)
famous paper stating that the amount of DNA varied with the species but always
preserving certain base ratios, and, finally, Hershey and Chase’s (1952) article
showing that only the DNA of phages entered the bacterial cell at the time of infection,
whereas the phage coat was left outside.
Notice that, by mere Bayesian updating, the probability that the transforming
agent was DNA increased with each of the tests; the probability that it was a
combination of DNA plus protein, on the other hand, increased as well, but not as
much. Granted, many of the aforementioned findings on genes and DNA did not make
explicit connections with prior experiments on bacterial transformation; however, once
we couple such results with Mirsky’s early understanding of transformation as a type
of genetic phenomenon – which, as we have seen, he was proud to acknowledge – then
the conclusion should be that they boosted H6’s prior more than that of H5. Thus,
PM(H6/ei)–PM(H6) > PM(H5/ei)–PM(H5), where ei refers, in each case, to the evidence
provided by the successive tests just mentioned. In short, even though we cannot
guarantee that Mirsky already held PM(H6) > PM(H5) by the late forties – this might
Eleonora Cresto – SHPS C (2008)
35
not have been the case if the initial difference between PM(H6) and PM(H5) had been
very large – this is indeed quite a plausible statement. (Of course, the later the time we
consider, the more likely that H6 obtains greater probability.)
Let me turn now to a possible reconstruction of Mirsky’s assessment of how
virtuous each of the two hypotheses really was, comparatively speaking. Consider first
how each of them ranked with respect to unification power. As I have already
mentioned, the extent to which a given hypothesis can be seen to enhance the
unification of a particular belief corpus depends, among other things, on the prior
contents of that corpus. Thus, in the previous section I have contented that, from Avery
and McCarty’s point of view, H5 hinted at a more complex picture of bacterial
transformation, and hence it should rank worst as far as unification power goes.
Clearly, this is not the case for Mirsky. Mirsky had devoted a large portion of his
career so far to the study of the behavior of proteins, and, in particular, since the early
forties he concentrated on nucleoproteins (cf. for instance Cohen 1998). If there was
somebody convinced of the subtleties proteins were able to display, it was him.
Accordingly, we should expect his set of full beliefs KM to differ from Avery’s or
McCarty’s in many crucial ways. In Mirsky’s case, the apparent complication of
referring to two substances rather than one (as responsible of the transformation) can
be outweighed by the fact that, by doing so, H5 helped Mirsky make a coherent picture
out of the received view on proteins and nucleic acids, together with the new findings.
In the light of this, it is reasonable to think that, for Mirsky, only H5 carried maximum
value for unification power.
Eleonora Cresto – SHPS C (2008)
36
On the other hand, Mirsky might have taken H6 to be the most testable and
economical, for reasons analogous to those already advanced in connection with
McCarty’s IBE: namely, that it was uncertain how to test the presence of substances
that could not be detected by state-of-the-art techniques. In addition, H5 and H6 might
have been perceived as equally accurate, in the sense that none of them described the
chemical composition of the transformation phenomenon at a substantively greater
level of detail.
Based on these considerations, we can conjecture that Mirsky proceeds as if
there exists some epistemic utility function fM, according to which unification power is
given more weight than other virtues, and which yields that EEU(Hi) = PM(Hi)fM(Hi) is
tied for H5 and H6 – even under the assumption that PM(H6) > PM(H5).
10. Conclusions
I hope to have shown that the model sketched in Section 3 can be successfully used to
clarify a number of misunderstandings about Avery’s attitude towards his own
research, as well as about the early reaction of the scientific community to the 1944
article. By being sensitive to the structure of questions and explanatory answers
presupposed in Avery’s case, the model helped us identify in what sense (if any) we
can say that Avery was overly cautious, and why. In addition, by being sensitive to the
structure of personal probabilities and epistemic virtues of different agents, the model
helped us understand why it was rational for both Mirsky and McCarty to proceed the
way they did, in spite of the fact that McCarty was an early supporter of the hypothesis
Eleonora Cresto – SHPS C (2008)
37
that the pneumococcal transforming principle was DNA, whereas Mirsky was reluctant
to accept it as late as in the early fifties.
As I can see it, the proposal sketched in this paper has clear advantages over
other models of inquiry. To begin with, many rival accounts simply do not apply to our
case study. Strict Hypothetico-Deductivism and Falsificationism are not adequate
contenders, since all interesting rival hypotheses were actually compatible with the
evidence. In addition, as revealed by many of the quotes from McCarty or Boivin (as
stated in section 8) agents sometimes do accept, or come to believe, non-tautological
statements, in the sense that they become truly convinced that such statements are
true. 20 No model of research can properly represent this situation unless it allows that
agents “detach” the conclusions of their non-deductive arguments and add them to
their epistemic corpora. These considerations count against positions such as radical
probabilism, as well as against traditional brands of Bayesianism, according to which
we are only entitled to accept evidential statements, but not general hypotheses.
As for other models of IBE, some of them do not construe IBE as an
acceptance rule at all, and hence, as with Bayesianism, they are unable to explain why
it might have been rational for scientists to fully believe that the transforming principle
was DNA. To mention a few well known examples, Day & Kinkaid (1994) conceive
of IBE as a highly contextual principle that helps us obtain the relevant priors and
likelihoods that feed Bayes’s theorem at the time of calculating the posterior
20
In this paper I have not distinguished between accepting a statement and fully believing it. For well
known proposals that trade on this distinction, cf. van Fraassen (1989, 2002), Cohen (1992), Maher
(1993), or Lehrer (2000). Other interesting suggestions can be found in Tuomela (2000), or Da Costa &
French (2003). For an overview of the treatment this subject has received in the literature, see the
articles in Engel (2000).
Eleonora Cresto – SHPS C (2008)
38
probability of hypotheses, via Bayesian conditionalization (see in particular ibid., pp.
285-286). Similarly, Okasha (2000) urges us to think of IBE as a rule that gives us “a
way of determining priors and likelihoods” (ibid., p. 703). Finally, in Lipton’s account
(as presented in Lipton, 2004) we find once again an ambiguous stance towards the
notion of IBE as an ampliative rule. Right after asserting that he takes IBE to deal with
hypothesis acceptance, he points out that he will “leave the notion of acceptance to one
side” in order to show that IBE and Bayesianism are actually compatible (ibid., p.
113). However, Lipton cannot show that this is indeed the case without betraying his
own proposal. Clearly, if an agent ever comes to believe new (non-evidential)
hypotheses, her so coming to believe them cannot be the result of conditionalizing on
the evidence: a conception of IBE as a tool that puts conditionalization in motion is
incompatible with a conception of IBE as an ampliative rule. Independently of this
problem, I am not sure whether a Lipton-style approach could give us a fine-grained
analysis as to why McCarty and Mirsky were entitled to carry out incompatible IBEs.
Within Lipton’s proposal, we could point out that the two scientists disagreed on
which hypothesis was the “loveliest” – where the loveliest hypothesis, in his jargon, is
the one that provides the most understanding (ibid., Ch. 4). But this is less informative
than what we intuitively demand from an explanation of the fact that McCarty and
Mirsky exhibited strikingly different epistemic attitudes.
As a final comment, let me address two possible sources of concern. In the first
place, it could be objected that the present model is strictly individualistic, and hence
that it is not sensitive to the way in which different groups of people interacted with
Eleonora Cresto – SHPS C (2008)
39
each other. 21 Against this objection, it should be pointed out that, within the present
framework, the epistemic state of an individual agent is allowed to change as a
response to suggestions or evidence provided by her colleagues; hence the fabric of a
particular belief state is in itself the result of an interactive process. In other words, the
fact that we aim at reconstructing individual perspectives and decisions does not force
us to eliminate references to social interaction. It is true, however, that the analysis
proposed here did not focus on the evolution of the scientific community as a whole.
But we can concede this and still claim that the model I favor constitutes exactly the
type of proposal we need in order to deal with (1), (2) and (5) from section 2 – to wit,
in order to clarify questions about Avery’s caution, about the timing of acceptance of
H6 by particular scientists, or about the rationality of incompatible epistemic attitudes.
The second potential concern is related to the fact that the analysis I have just
offered assumed all along that the attitudes of Avery, McCarty or Mirsk had a rational
core. I believe the charitable procedure paid off, in the sense that it enabled us to
obtain a richer mental picture of the case, from an epistemic point of view. It is worth
stressing, however, that I do not think rational reconstructions are always desirable; I
do not think that we should always attempt to provide explanations (for a given pattern
of behavior) that grant the rationality of the agents involved. Indeed, nothing prevents
us from using the given framework to actually distinguish rational from irrational
epistemic attitudes. This is, of course, as it should be. No abstract model should ever
force us to violate our basic intuitions as to whether a concrete case in the history of
science is an instance of good, or not-so-good, epistemic behavior. Rationality is in the
21
I want to thank an anonymous referee for putting forward this objection.
Eleonora Cresto – SHPS C (2008)
40
eye of the beholder, so to speak. All a model should do is be flexible enough so as to
embody and refine our prior intuitions in the best possible terms – as well as tell us
how to proceed when no clear intuitions are at stake. And, in this sense, the proposal
suggested in Section 3 delivers just what we hoped to obtain.
Acknowledgments
I would like to thank John Collins, Philip Kitcher, Isaac Levi, Achille Varzi, and an
anonymous referee for Studies in History and Philosophy of Biological and
Biomedical Sciences for very helpful comments and suggestions.
References
Alloway, J. L. (1932). The transformation in vitro of R pneumococci into S forms of
different specific types by the use of filtered pneumococcus extracts. Journal of
Expermiental Medicine, 55, 91-99.
Alloway, J. L. (1933). Further observations on the use of pneumococcus extracts in
effecting transformation of type in vitro. Journal of Experimental Medicine, 57, 265278.
Amsterdamka, O. (1993). From pneumonia to DNA: the research career of Oswald T.
Avery. Historical Studies in the Physical and Biological Sciences, 24, 1-40.
Avery, O. T. (1947). Report of Dr. Avery (assisted by Drs. Hotchkiss, McCarty and
Taylor). In Report of the Director of the Hospital to the Corporation of the Rockefeller
Institute for Medical Research (pp. 126–136). 19 April. Sleepy Hollow, NY,
Rockefeller Archive Center. (Available at
http://profiles.nlm.nih.gov/CC/A/A/N/R/_/ccaanr.pdf)
Eleonora Cresto – SHPS C (2008)
41
Avery, O. T., & Heidelberg, M. (1923). Immunological relationships of cell
constituents of pneumococcus. Journal of Experimental Medicine, 38, 81.
Avery, O. T., & Horsfall, F. L. (1943). Report of Drs. Avery and Horsfall: Study on
the chemical nature of the substance inducing transformation of specific types of
pneumococcus (Avery and McCarty). Scientific Report to the Corporation and the
Board of Scientific Directors of the Research Institute, 31 (1942–1943), 143–175.
Sleepy Hollow, NY, Rockefeller Archive Center. (Available at
http://profiles.nlm.nih.gov/CC/A/A/D/S/_/ccaads.pdf)
Avery, O. T., MacLeod, C. M., & McCarty, M. (1959). Studies on the chemical nature
of the substance inducing transformation of pneumococcal types: Induction of
transformation by a deoxyribonucleic acid fraction isolated from pneumococcus type
III. In J. A. Peters (Ed.), Classic papers in genetics (pp. 173–192). Englewood Cliffs,
NJ: Prentice Hall. (First published in 1944, Journal of Experimental Medicine, 79,
137–158).
Beurton, P. J., Falk, R., & Rheinberger, H.-J. (Eds.). (2000). The concept of the gene in
development and evolution. Cambridge: Cambridge University Press.
Bilgrami, A. (2000). Is Truth a Goal of Inquiry? Rorty and Davidson on Truth. In R.
Brandon (Ed.), Rorty and his critics (pp. 242-261). Oxford: Blackwell.
Bilgrami, A. (2004). Skepticism and Pragmatism. In D. McManus (Ed.), Wittgenstein
and scepticism (pp. 56-75). London: Routledge.
Boivin, A. (1947). Directed mutation in colon bacilli, by an inducing principle of
deoxyribonucleic nature: Its meaning for the general biochemistry of heredity. Cold
Spring Harbor Symposium on Quantitative Biology, 12, 7-17.
Boivin, A., Vendrely, R., & Vendrely, C. (1948). L’acide desoxyribonucleique du
noyau cellulaire, depositaire des caractéres hereditaires; arguments d’ordre analytique.
Comptes Rendus Hebdomadaires des Séances de l’Académie des Sciences, 226, 10611063.
Burian, R. M. (1985). On conceptual change in biology: the case of the gene. In D. J.
Depew, & B. H. Weber (Eds.), Evolution at a crossroads: the new biology and the new
philosophy of science (pp. 21-42). Cambridge: The MIT Press.
Burnet, F. M. (1968). Changing patterns: an atypical autobiography. Melbourne:
Heinemann.
Cairns, J., Stent, G., & Watson, J. (Eds.). (1966). Phage and the origins of molecular
biology. Cold Spring Harbor, Long Island: Cold Spring Harbor Laboratory of
Molecular Biology.
Eleonora Cresto – SHPS C (2008)
42
Carlson, E. A. (1989). The gene: a critical history. Ames, Iowa: Iowa State University
Press. (First published 1966)
Chadwick, J. (1932). Possible existence of a neutron. Nature, 129, 312.
Chargaff, E. (1950). Chemical specificity of nucleic acids and mechanism of their
enzymatic degradation. Experientia, 6, 201–240.
Chargaff, E. (1978). Heraclitean fire: sketches from a life before nature. New York:
The Rockefeller University Press.
Chargaff, E. (1979). How genetics got a chemical education. Annals of the New York
Academy of Sciences, 325, 345-360.
Cohen, J. (1992). An essay on belief and acceptance. Oxford: Clarendon Press.
Cohen, S. S. (1998). Alfred Ezra Mirsky. Biographical memoirs. National Academy of
Sciences, 73, 322-332.
Cresto, E. (2006). Inferring to the best explanation: a decision-theoretic approach.
Ph.D. thesis, Columbia University, New York.
Da Costa, N. C. A., & French, S. (2003). Science and partial truth: A unitary
approach to models and scientific reasoning. New York: Oxford University Press.
Dawes, H. (2004). The quiet revolution. Current Biology, 14, R605-R607.
Dawson, M. H., & Sia, R. H. P. (1931). In vitro transformation of pneumococcal types.
I. A technique for inducing transformation of pneumococcal types in vitro. Journal of
Experimental Medicine, 54, 681-700.
Day, T., & Kincaid, H. (1994). Putting inference to the best explanation in its place.
Synthèse, 98, 271-295.
Deichmann, U. (2004). Early responses to Avery et al.'s paper on DNA as hereditary
material. Historical Studies in the Physical and Biological Sciences, 34, 207-232.
Diamond, A. (1982). Avery’s ‘neurotic reluctance’. Perspectives in Biology and
Medicine, 26, 132-136.
Dubos, R. J. (1976). The Professor, the institute and DNA. New York: The Rockefeller
University Press.
Engel, P. (Ed.) (2000). Believing and accepting. Dordrecht: Kluwer.
Eleonora Cresto – SHPS C (2008)
43
Fleming, D. (1968). Emigré physicists and the biological revolution. Perspectives in
American History, 2, 152-189.
Fuhrmann, A. (1997). An essay on contraction. Stanford, CA: Center for the Study of
Language and Information.
Griffith, F. (1928). The significance of pneumococcal types. Journal of Hygiene, 27,
113-159.
Heidelberg, M., & Avery, O. T. (1923). The soluble specific substance of
pneumococcus. Journal of Experimental Medicine, 40, 301.
Hershey, A. D., & Chase, M. (1952). Independent functions of viral protein and
nucleic acid in growth of bacteriophage. Journal of General Physiology, 36, 39-56.
Hotchkiss, R. (1949). Études chimiques sur le facteur transformant du pneumocoque.
In Centre National de la Recherche Scientifique, Unités biologiques douées de
continuité génétique. Paris, Juin–Juillet 1948 (pp. 57–65). Colloques Internationaux
du C. N. R. S., 8. Paris: Centre National de la Recherche Scientifique.
Hotchkiss, R. (1952). The role of desoxyribonucleates in bacterial transformation. In
W. D. McElroy, & B. Glass (Eds.), Phosphorus metabolism. A symposium on the role
of phosphorus in the metabolism of plants and animals, vol. II (pp. 426-439).
Baltimore: John Hopkins Press.
Hotchkiss, R. (1965). Oswald T. Avery. Genetics, 51, 1-10.
Hotchkiss, R. (1966). Gene, transforming principle and DNA. In J. Cairns, G. S. Stent,
& J. D. Watson (Eds.), Phage and the origins of molecular biology (pp. 180-200).
Cold Spring Harbor, Long Island: Cold Spring Harbor Laboratory of Molecular
Biology.
Hotchkiss, R. (1979). The identification of nucleic acid as genetic determinants.
Annals of the New York Academy of Sciences, 325, 321-342.
Judson, H. (1980). Reflections on the historiography of molecular biology. Minerva,
18, 369-421.
Judson, H. (1996). The eighth day of creation: the makers of the revolution in biology.
New York: Simon & Schuster. (First published 1978)
Kitcher, P. (1982). Genes. British Journal for the Philosophy of Science, 33, 337-359.
Kitcher, P. (1984). 1953 and all that: a tale of two sciences. Reprinted in P. Kitcher
(2003), In Mendel’s mirror: philosophical reflections on biology (pp. 3-30). New
York: Oxford University Press.
Eleonora Cresto – SHPS C (2008)
44
Kitcher, P. (1992). Gene: current usages. In E. Keller, & E. Lloyd (Eds.), Keywords in
evolutionary biology (pp. 128-131). Cambridge, Mass.: Harvard University Press.
Kuhn, T. (1962). The structure of scientific revolutions. (2nd ed.). Chicago: University
of Chicago Press. (First published 1962).
Lederberg, J. (1972). Reply to H.V. Wyatt. Nature, 239, 234.
Lederberg, J. (2000). The dawning of molecular genetics. Trends in microbiology, 8,
194-5.
Lederberg, J., & Tatum, E. L. (1946). Gene recombination in Escherichia coli. Nature,
158, 558.
Lehrer, K. (2000). Theory of knowledge. Boulder: Westview Press.
Levene, P., & Bass, L. W. (1931). Nucleic acids. American Chemical Society
Monograph Series. New York: Chemical Catalog Company.
Levi, I. (1980). The enterprise of knowledge. Cambridge, Mass: The MIT Press.
Levi, I. (1984). Decisions and revisions. Cambridge: Cambridge University Press.
Levi, I. (1994). How to fix a prior. In D. Prawitz, & D. Westerstahl (Eds.), Logic and
philosophy of science in Uppsala (pp. 185-204). Dordrecht: Kluwer.
Levi, I. (1997). The covenant of reason. Cambridge: Cambridge University Press.
Lipton, P. (2004). Inference to the best explanation (2nd ed.). London: Routledge. (First
published 1991)
Luria, S. E., & Delbrück, M. (1943). Mutations of bacteria from virus sensitivity to
virus resistance. Genetics, 28, 491-511.
Maher, P. (1993). Betting on theories. Cambridge: Cambridge University Press.
Maienschein, J. (1992). Gene: historical perspectives. E. Keller, & E. Lloyd (Eds.),
Keywords in evolutionary biology (pp. 122-127). Cambridge, Mass.: Harvard
University Press.
McCarty, M. (1946). Chemical nature and biological specificity of the substance
inducing transformation of pneumococcal types. Bacteriological Reviews, 10, 63-71.
McCarty, M. (1985). The transforming principle: discovering that genes are made of
DNA. New York: Norton.
Eleonora Cresto – SHPS C (2008)
45
McCarty, M. (1994). A retrospective look: how we identified the pneumococcal
transforming substance as DNA. Journal of Experimental Medicine, 179, 381-394.
McCarty, M., & Avery, O. T. (1946a). Studies on the chemical nature of the substance
inducing transformation of pneumococcal types. Part II: effect of the
desoxyribonuclease on the biological activity of the transforming substance. Journal of
Experimental Medicine, 83, 89-96.
McCarty, M., & Avery, O. T. (1946b). Studies on the chemical nature of the substance
inducing transformation of pneumococcal types. Part III: an improved method for the
isolation of the transforming substance and its application to the pneumoccocus types
II, III and VI. Journal of Experimental Medicine, 83, 97-104.
Mirsky, A. (1947). Chemical properties of isolated chromosomes. Cold Spring Harbor
Symposium on Quantitative Biology, 12, 143-146.
Mirsky, A. (1951). Some chemical aspects of the cell nucleus. In L. C. Dunn, Genetics
in the 20th century: essays on the progress of genetics during its first 50 years. New
York: Macmillan.
Mirsky, A. (1972). Letter to Joshua Lederberg. 31 October. Sleepy Hollow, NY,
Rockefeller Archive Center. (Available at
http://profiles.nlm.nih.gov/CC/A/A/H/P/_/ccaahp.pdf)
Mirsky, A. (1973). Letter to Jack S. Cohen. 29 June. Sleepy Hollow, NY, Rockefeller
Archive Center. (Available at http://profiles.nlm.nih.gov/CC/A/A/H/R/_/ccaahr.pdf).
Mirsky, A., & Pollister, W. (1946). Chromosin, a desosyribose nucleoprotein complex
of the cell nucleus. Journal of General Physiology, 30, 117-148.
Mirsky, A., & Ris, H. (1949). Variable and constant components of chromosomes.
Nature, 163, 666-667.
Okasha, S. (2000). Van Fraassen’s critique of inference to the best explanation. Studies
in History and Philosophy of Science, 31, 691-710.
Olby, R. (1990). The molecular revolution in biology. In R. C. Olby, G. N. Cantor, J.
R. R. Christie, & M. J. S. Hodge (Eds.). Companion to the history of modern science
(pp. 503-520). London: Routledge.
Olby, R. (1994). The path to the double helix: the discovery of DNA (2nd ed.). New
York: Dover. (First published 1974)
Peirce, C. S. (1931-1958). In C. Harstshorne, P. Weiss, & A. Burks (Eds.). Collected
papers of Charles Sanders Peirce (8 vols.) Cambridge, MA: Harvard University Press.
Eleonora Cresto – SHPS C (2008)
46
Perutz, M. (1994). Letter to Joshua Lederberg. 31 March. Sleepy Hollow, NY,
Rockefeller Archive Center. (Available at
http://profiles.nlm.nih.gov/CC/A/A/I/E/_/ccaaie.pdf)
Peters, J. A. (Ed.). (1959). Classic papers in genetics. Englewood Cliffs, N.J.:
Prentice-Hall.
Pollock, M. (1970). The discovery of DNA: an ironic tale of chance, prejudice and
insight. Journal of General Microbiology, 63, 1-20.
Russell, N. (1988). Oswald Avery and the origin of molecular biology. British Journal
for the History of Science, 21, 393-400.
Shimony, A. (1970). Scientific inference. Reprinted in A. Shimony (1993), Search for
a Naturalistic World View. Vol. I: Scientific Method and Epistemology (pp. 183-273).
Cambridge: Cambridge University Press.
Stent, G. S. (1972). Prematurity and uniqueness in scientific discovery. Scientific
American, 227, 84-93.
Toulmin, S. (1972). Human understanding (Vol. 1). Princeton, N. J.: Princeton
University Press.
Tuomela, R. (2000). Belief versus Acceptance. Philosophical Explorations, 2, 22-157.
Van Fraassen, B. (1989). Laws and symmetry. Oxford: Clarendon Press.
Van Fraassen, B. (2002). The empirical stance. New Haven: Yale University Press.
Watson, J., & Berry, A. (2003). DNA: the secret of life. New York: Alfred A. Knopf.
Wyatt, V. (1972). When does information become knowledge? Nature, 235, 86-89.