on the shoulders of robert merton: towards a normative theory of citation

9
Jointly published by AkadØmiai Kiad, Budapest Scientometrics, and Kluwer Academic Publishers, Dordrecht Vol. 60, No. 1 (2004) 7179 Received February 4, 2004 Address for correspondence: HENRY SMALL Thomson ISI, 3501 Market Street, Philadelphia, PA 19104, USA E-mail: [email protected] 01389130/2004/US $ 20.00 Copyright ' 2004 AkadØmiai Kiad, Budapest All rights reserved On the shoulders of Robert Merton: Towards a normative theory of citation HENRY SMALL Thomson ISI, Philadelphia, PA (USA) In a series of seminal studies Robert K. Merton created a coherent theoretical view of the social system of science that includes the salient features of the formal publication system, thereby providing a theoretical basis for scientometrics and citationology. A fundamental precept of this system is the view of citations as symbolic payment of intellectual debts. When this concept is merged with a complementary theory of the conceptual symbolism of citations, the possibility for a rapprochement of the normative and constructivist theories is achieved, where the dual function of citations as vehicles of peer recognition and constructed symbols for specific original achievements in science is realized. This new synthesis is embodied in a citation classification system, the citation cube, with dimensions of normative compliance, symbolic consensus, and disinterestedness (self-citation). I cannot claim to have known Robert Merton well or the full range of his scholarship. But the encounters I had with him over the years, both in person and in reading his work, were memorable and intellectually decisive. He was one of those unique and original geniuses who could make each moment you spent with him an intense intellectual experience. I recall one such encounter in the early 1970s when I was struggling with the methodological decision of whether to focus my energies on co- words or co-citations as the structural building blocks for science. The topic of discussion was his memoir On the Shoulders of Giants (MERTON, 1965) and it occurred to me that the use of the plural of giant suggested that a scientist could in effect be standing on the shoulders of many giants, and that the aggregate pattern of giant straddling was equivalent to the co-citation idea I had been toying with. The further implication was that the resulting groups of giants rubbing shoulders, as it were, would be a structure of the co-allocation of rewards, i.e., giants who are being rewarded for achievements perceived to be related, at least by those doing the straddling. Thus, seen in this way the structure of science is generated by patterns of co-recognition. I am sure that this insight strengthened my resolve to pursue the co-citation rather than the co-word route (SMALL, 2003), and led me to the work of Gene Garfield and the field of citationology (GARFIELD, 1998). Later on, reading Merton on originality and priority in science, I realized how these norms could be expressed in the choices scientists make

Upload: henry-small

Post on 06-Aug-2016

217 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: On the shoulders of Robert Merton: Towards a normative theory of citation

Jointly published by Akadémiai Kiadó, Budapest Scientometrics,and Kluwer Academic Publishers, Dordrecht Vol. 60, No. 1 (2004) 71–79

Received February 4, 2004Address for correspondence:HENRY SMALLThomson ISI, 3501 Market Street, Philadelphia, PA 19104, USAE-mail: [email protected]

0138–9130/2004/US $ 20.00Copyright © 2004 Akadémiai Kiadó, BudapestAll rights reserved

On the shoulders of Robert Merton:Towards a normative theory of citation

HENRY SMALL

Thomson ISI, Philadelphia, PA (USA)

In a series of seminal studies Robert K. Merton created a coherent theoretical view of thesocial system of science that includes the salient features of the formal publication system, therebyproviding a theoretical basis for scientometrics and citationology. A fundamental precept of thissystem is the view of citations as symbolic payment of intellectual debts. When this concept ismerged with a complementary theory of the conceptual symbolism of citations, the possibility fora rapprochement of the normative and constructivist theories is achieved, where the dual functionof citations as vehicles of peer recognition and constructed symbols for specific originalachievements in science is realized. This new synthesis is embodied in a citation classificationsystem, the citation cube, with dimensions of normative compliance, symbolic consensus, anddisinterestedness (self-citation).

I cannot claim to have known Robert Merton well or the full range of hisscholarship. But the encounters I had with him over the years, both in person and inreading his work, were memorable and intellectually decisive. He was one of thoseunique and original geniuses who could make each moment you spent with him anintense intellectual experience. I recall one such encounter in the early 1970s when Iwas struggling with the methodological decision of whether to focus my energies on co-words or co-citations as the structural building blocks for science. The topic ofdiscussion was his memoir “On the Shoulders of Giants” (MERTON, 1965) and itoccurred to me that the use of the plural of giant suggested that a scientist could ineffect be standing on the shoulders of many giants, and that the aggregate pattern ofgiant straddling was equivalent to the co-citation idea I had been toying with. Thefurther implication was that the resulting groups of giants rubbing shoulders, as it were,would be a structure of the co-allocation of rewards, i.e., giants who are being rewardedfor achievements perceived to be related, at least by those doing the straddling. Thus,seen in this way the structure of science is generated by patterns of co-recognition. I amsure that this insight strengthened my resolve to pursue the co-citation rather than theco-word route (SMALL, 2003), and led me to the work of Gene Garfield and the field ofcitationology (GARFIELD, 1998). Later on, reading Merton on originality and priority inscience, I realized how these norms could be expressed in the choices scientists make

Page 2: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

72 Scientometrics 60 (2004)

when citing the literature. When scientists agree on what constitutes prior relevantliterature, including what is significant in that literature, they are in fact defining thestructures of their communities.

Even though he did not himself labor in the vineyards of citation analysis, inretrospect Merton’s sociology of science provides the most coherent theoreticalframework available for this enterprise. In fact, one would be tempted to call him acrypto-citationologist since so much of his work on the rewards system of science restsexplicitly or implicitly on the system of publication and citation in science. Of course,many of his students have extended his work in this direction. But rather than reviewthe work of these intellectual offspring, I will focus on the progenitor, and attempt toproject his ideas through the lens of my own research agenda, perhaps distorting orbending his views, with apologies to the original.

I will admit that in my case I was late to fully appreciate my indebtedness to hisideas, and was initially deflected from the co-recognition interpretation of co-citation bymy interest in finding a parallel between co-citations and informal communicationbetween scientists. However, over the years I came to see the limitations of that view(BALDI, 1998) and the correctness of Merton’s notions of citations as atoms of peerrecognition (MERTON, 1988). Another source of resistance was my interest in citationsas a symbolic language for science, which could link directly with the substance of theunderlying science (SMALL, 1978).

Belver Griffith, my late collaborator on mapping science, was fond of saying thatthis or that sociologist of science, mainly of the constructivist persuasion, was in needof a strong dose of Robert Merton. His prescription was: take two Mertons and see mein the morning (GRIFFITH, 1997). One could argue that in jettisoning Mertonian notionssuch as the rewards system of science, the norms of science, the quest of priority, theMatthew effect, and the concept of multiples, constructivists have missed manyopportunities for research. Indeed the normative and constructivist approaches haveseemed at times to be as at odds with one another as the wave and particleinterpretations in quantum physics. It is easy to caricature norms as rigid legalistic rules,just as it is to parody the ideas of constructivists that physical reality is sociallyconstructed. However, there is no doubt that norms, however you regard their force onscientific behavior, have provided a solid basis for studies of the publication system ofscience, and a set of expectations for conventional scientific behavior against which wecan measure deviations.

To review some implications of Merton’s thought we begin with the drive to publishin science. Unlike the case of proprietary knowledge, he stressed that scientists aremotivated to openly disclose their findings in the form of publication: scientists giveaway their ideas in order to claim ownership of them (MERTON, 1957). They do so toestablish a claim to priority that carries the legitimacy of a reputable peer reviewedjournal and a time stamp of publication date that can establish that they were the first.

Page 3: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

Scientometrics 60 (2004) 73

Secondly they are motivated by the hope and expectation that their work will beacknowledged by their peers, not the least in the form of future citations to it. So inexchange for their freely offered contributions, the scientific community provides theopportunity for recognition. Through this we understand why the invention of thejournal was so crucial in providing a mechanism for deciding priority and ascribingcredit.

We might add that scientists are motivated to reference the work of their peers, inpart because they believe in the justice of giving credit where credit is due, but also as akind of mortgaging of their present value for future income which they hope to receivethrough peer recognition. In other words, giving credit to others may diminish yourclaim to originality but increases the likelihood that you will receive the attention ofothers. This creates a kind of golden rule of scholarship – to receive credit you mustgive credit to others, and in the process both donor and recipient are validated(MERTON, 1969). Thus the system of publication and citation feeds upon itself.Undoubtedly the success of this system is due at least in part to the fact that it is self-referential, unlike achievements in the world of art, music or literature that rely more onthe external validation of critics or the general public, as well as peers (CRANE, 1976).

The norm of communism (MERTON, 1942), or as he later preferred, thecommunality of scientific knowledge, is in sharp focus today as authors pressurepublishers for open access to their published work via the web, without the impedimentof copyright. This norm asserts the common ownership of scientific knowledge and byextension of scientific publications, the conveyors of that knowledge. The norm ofcommunality views science as a worldwide collaborative effort, with scientists buildingon each other’s work. To the degree that there is secrecy or withholding of information,this collaboration is diminished, and the rate of discovery retarded.

But such cooperation is possible only because the conventions of publication allowfor the orderly disposition of knowledge claims and the just allocation of credit forcontributions. The altruistic and communal spirit is driven by the self-interest of thescientist in gaining priority and symbolic recognition for his or her work. Discovery,priority and recognition are thus inextricably intertwined with publication, and citationlinks expose the socially validated structure of originality. This may explain whycitation indexes are so popular and natural for scientists to use for information retrieval.By revealing how authors allocate credit to predecessors and have themselves beencredited, citation indexes hold up a mirror to the rewards system of science, which inturn helps authors find appropriate works to read and cite, thus feeding back into therewards system.

Merton’s work also brought the struggle for priority into sharp focus by numerousexamples drawn from history (MERTON, 1957). His work on multiples (MERTON, 1961)was an extension of this struggle for priority in showing how often scientistsfind themselves engaged in essentially the same work as their colleagues. Research in

Page 4: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

74 Scientometrics 60 (2004)

co-citation analysis reinforces the importance of multiples because clusters of co-citedpapers often explicitly represent simultaneous, independent co-discoveries, such as theexample of Baltimore and Temin’s independent discovery of reverse transcriptase thatemerged in one of our first co-citation experiments (SMALL & GRIFFITH, 1974). Butsuch widely acknowledged co-discoveries were only the tip of the iceberg. Mertonclearly had a broader view of what multiples might be, extended in both geography andtime, with varying degrees of conscious or unconscious duplication (MERTON, 1963a).Thus I have often wondered whether his multiples, construed in their widest possiblesense as groups of topically related papers, published at different times, bygeographically dispersed authors, who partially duplicate each others’ findings are infact the equivalent of co-citation clusters.

In the citation world, his Matthew Effect (MERTON, 1968) has a clear interpretation.When a paper is cited, other authors can see that it is, and this heightens their interest inthe paper and the likelihood of their citing it as well at some later date. In this sense,citation acts like an expert referral. The operation of this feedback mechanism, once setin motion will increase the inequality of citations by focusing attention on a smallernumber of selected sources, and widening the gap between symbolically rich and poor.The cycle may be broken by the appearance of new work that captures the attentionfrom the older source, effectively replacing it. The Matthew effect is manifest at thedocument level by the skewed distribution of cites to documents (PRICE, 1965), roughlyfollowing a power law. This is not to say that the eminence of an author, journal orinstitution does not contribute to the propensity of a document to be cited, the so-calledhalo effect (COLE & COLE, 1973 p.221). But the most significant determinant of citationmay have to due with priority and originality, and, at a yet deeper level, the consistencyof the contribution with current “canonical knowledge” (STENT, 1972). Similarlymultiples must also be somehow related to consistency with current knowledge whichsets the stage for many to realize the same thing at nearly the same time.

The importance of citation was acknowledged by MERTON (1988) as “. . . what issurely the most widespread and altogether basic form of scholarly recognition, thatwhich comes with having one’s work used and explicitly acknowledged by one’speers.” But he further stated that such use of scientists’ work by their peers is only“crudely indexed by the number of citations to it.” The crudeness of this measure, wemight presume, derives from many factors that run counter to his norms and from theopen-ended nature of citations and their primary function of pointing to sources. Self-citation, for example, would appear on the surface to violate the norm of humility eventhough it legitimately points the reader to a valid source. Negative citations, althoughthey may represent important influences (COLE & COLE, 1973, p. 25), are not of thesame order as positive citations. His own concept of obliteration by incorporation (OBI)demonstrated how normatively valid citations could be lost (MERTON, 1979). On theother side of the coin, there is the over recognition of some work due to the operation of

Page 5: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

Scientometrics 60 (2004) 75

the Matthew effect. These and other factors made Merton and others dubious about theuse of simple citation counts for evaluative purposes (MERTON, 1977, p. 53), while atthe same time welcoming citation indexing as a new tool for sociological analysis.

In discussing the reward system of science, he gave his main attention to the ways inwhich individual scientists receive recognition, such as awards, eponymy, and historicalmentions (MERTON, 1957). However, it seems evident that the main target ofrecognition is the discrete contribution to knowledge in the form of the openpublication. In this view, credit or discredit flows first to the publication, and then to theassociated authors, institutions, countries, and journals that are its attributes. Theindividual scientist derives prestige by a summation of recognition bestowed on his orher published works, which may vary widely in importance. Such a document-centricview of the rewards system is perhaps one that Merton would not have beencomfortable with. Here the published artifact takes center stage, with the authorassuming a supporting or facilitating role.

This is not to suggest that somehow publications mysteriously appear without theefforts of individual authors, or that authors do not bear full responsibility for theirproducts. But as the solitary author is replaced by a collaborative team (BIAGIOLI &GALISON, 2003), the image of the heroic individual at the lab bench becomes moreproblematic, as Merton pointed out early on (1963b). While the apportioning of creditbecomes more difficult, this distancing of the publication from its author may have thedesirable effect of allowing work to be judged in a more objective and impersonalmanner, consistent with the norms of universalism and disinterestedness. Certainly afterpublication the published work can take on a life of its own independent of its author(SMALL, 1984) similar to how we view a work of art after its creator has completed it.

Building on Merton’s ideas, Kaplan posited a norm of citation, derived presumablyfrom the norms of originality and humility. In its purest sense such a norm might be theexpectation that authors acknowledge prior work in an accurate manner and true to theoriginal authors’ intentions. Accurate citation practice fosters the cumulative nature ofscience, requiring scholars to keep up with the literature. There are negative sanctionsfor authors who do not cite the appropriate sources or cite them inappropriately orinaccurately, although many such abuses may go unnoticed (PALEVITZ, 1997). Ofcourse, a norm of citation, like other norms, is not a rigid constraint. It does notprescribe how a work must be cited, and indeed the original author has little controlover this process. The citing author may transform the work into somethingunrecognizable to the cited author. It may be the case that prior literature is necessarilyin a constant state of reinterpretation, adapting to changes in knowledge within the field(COZZENS, 1988). This process can involve changes in emphasis, selection of content,reinterpretation or transformation. Such poetic license can be contrasted with deviantcases such as distortion, falsification, misallocation of credit, or failure to credit such asOBI. Such processes can lead to conflict among authors, but within certain limits the

Page 6: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

76 Scientometrics 60 (2004)

interpretative process may be an inevitable and healthy process, leading to arevitalization of earlier work as it is recast by new generations of researchers. Bycontrast LATOUR’s adversarial view of citation practice (1987), in which authors deployreferences in any way that will support their cause, is presumably not burdened by anynormative constraints. Nevertheless, a normatively directed constructive process is awork.

It is here that normative and constructivist approaches can find common ground. Ifthe norm of citation involves a symbolic payment of intellectual debt, it is, at the sametime, an ascription of meaning to the cited text and a construction of its meaning.Documents function as concept symbols for citing authors (SMALL, 1978) at the sametime symbolic credit or discredit is being bestowed. By extension, when documents areco-cited, citing authors are awarding co-recognition as well as creating an association ofmeanings. Here the symbolism of reward and the symbolism of meaning are operatingin tandem. For an individual document the idea for which the author is being creditedcan be congruent or at odds with the author’s original message or with consensual usagein the field. When the constructed meaning is coincident with the author’s originalmessage as well as common usage, we may say that there is strong normativecompliance. This would, I suspect, be the most common case. When the constructedand original messages diverge, we have what is for a normative sociologist deviantpractice on the part of the citing author, but perhaps normal behavior for aconstructivist. The research question for us is the frequency and degree of thisconvergence or divergence of meaning, and in the instances of divergence how theconflicting meanings are instrumental or useful to the citing authors sufficient to justifythe risk of sanction.

Nowadays there appears to be a reawakening of interest in formulating a theory ofcitation (CRONIN, 2000). Such a theory must encompass the spectrum of observedbehaviors from the most common forms such as ceremonial or perfunctory citation tothe less common deviant cases, such as negative citation, self-citation, andmisattribution. The empirical heart of such a theory is the comparison of the cited textwith its context of citation in the citing texts. One possible framework is to consider thedimensions of literalness versus consensus as shown in the figure. On the vertical axisliteralness measures the congruence of the cited work and the citing context from low tohigh. The horizontal axis labeled consensus indicates the degree of agreement in theciting community. Alternatively this axis might be labeled revolutionary versus normalscience.

Page 7: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

Scientometrics 60 (2004) 77

Consensus

Low HighLiteralness Low CONSTRUCTIVE

misattribution,reinterpretation,idiosyncratic negative,revolutionary

CONSTRUCTIVE

ritual negative,common misattribution,conventionaltransformation,obliteration byincorporation

High

NORMATIVE

substantive, organic,creative connection,unusual quotation

NORMATIVE

perfunctory, ceremonial,common direct quotation,conventionalinterpretation,paradigmatic

Figure 1. The figure shows two dimensions of the citation cube model, literalness and consensus, each brokeninto high and low categories. The third dimension of the cube, self-citation or disinterestedness, is not shown.The dimensions shown highlight the similarities or differences between the original cited text and the citingcontext, and whether these differences are widely adopted by the community or unique to the citing author.

For example, in this scheme negative citation would have low literalness since thecitation diverges from the original text, but could have low consensus if the author wasalone in his disagreement or high consensus if the citing author’s critical opinion wasshared by the community. In the latter case, the negatively cited work has becomesymbolic of an incorrect idea. Normatively compliant citations, such as perfunctory orceremonial citations (MORAVCSIK & MURUGESAN, 1975) would concentrate in the highliteral, high consensus box, while deviant cases, such as a revolutionary negativecitation or paradigm breaking reinterpretation, would fall mainly into the low literal,low consensus region. In general, Merton’s recognition model would work best for highliteral citations, while the constructivist model would work best for low literal citations.A mis-citation or misattribution could fall in either of the low literal boxes, dependingon its prevalence, but an OBI case would fit best in the high consensus, low literalcategory, since such citations are uniformly omitted. A particularly apt but notuniversally recognized citation, sometimes described as organic or substantive, would

Page 8: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

78 Scientometrics 60 (2004)

fall into the low consensus, high literal category. A commonly recognizedreinterpretation or reformulation would fall into the high consensus, low literalcategory, as would a commonly made misinterpretation or distortion. In this scheme,the degree of self-citation would constitute a dimension orthogonal to the first two,creating a citation cube consisting of eight categories. This third dimension might alsobe called the degree of disinterestedness, after one of Merton’s norms, because itreflects the degree to which the interests of the citing party are bound up with theinterests of the cited.

While such a mechanical scheme may seem remote from the elegance of Merton’sthought, it builds on the foundation he has provided, and when we get down topopulating this citation cube with actual cases, I believe that we will again reaffirm thepower of his norms in regulating citation behavior.

References

BALDI, S. (1998), Normative versus social constructivist processes in the allocation of citations: a network-analytic model. American Sociological Review, 63 (6) : 829–846.

BIAGIOLI, M., GALISON, P. (Eds) (2003), Scientific Authorship: Credit and Intellectual Property in Science.New York: Routledge.

COLE, J. R., COLE, S. (1973), Social Stratification in Science. Chicago: University of Chicago Press.COZZENS, S. E. (1988), Life-history of a knowledge claim: the opiate receptor case. Knowledge-Creation

Diffusion Utilization, 9 (4) : 511–529.CRANE, D. (1976), Reward systems in art, science and religion. American Behavioral Scientist,

19 (6) : 719–734.CRONIN, B. (2000), Semiotics and evaluative bibliometrics. Journal of Documentation, 56 (4) : 440–453.GARFIELD, E. (1998), Random thoughts on citationology, its theory and practice. Scientometrics,

43 (1) : 69–76.GRIFFITH, B. C. (1997), Discussion of nature of science provokes hit or miss debate. Physics Today,

50 (1) : 11.KAPLAN, N. (1965), The norms of citation behavior: prolegomena to the footnote. American Documentation,

16 (3) : 179–184.LATOUR, B. (1987), Science in Action. Cambridge, Mass.: Harvard University Press.MERTON, R. K. (1942), Science and technology in a democratic order. Journal of Legal and Political

Sociology, 1 : 115–126. Reprinted in: The Sociology of Science: Theoretical and EmpiricalInvestigations. Chicago: University of Chicago Press, 1973, pp. 267–278.

MERTON, R. K. (1957), Priorities in scientific discovery. American Sociological Review, 22 (6) : 635–659.Reprinted in: The Sociology of Science: Theoretical and Empirical Investigations. Chicago: University ofChicago Press, 1973, pp. 286–324.

MERTON, R. K. (1961), Singletons and multiples in scientific discovery. Proceedings of the AmericanPhilosophical Society, 105 (5) : 470–486. Reprinted in: The Sociology of Science: Theoretical andEmpirical Investigations. Chicago: University of Chicago Press, 1973, pp. 343–370.

MERTON, R. K. (1963a), Resistance to the systematic study of multiple discoveries in science. EuropeanJournal of Sociology, 4 : 237–249. Reprinted in: The Sociology of Science: Theoretical and EmpiricalInvestigations. Chicago: University of Chicago Press, 1973, pp. 371–382.

MERTON, R. K. (1963b), Resistance to the systematic study of multiple discoveries in science. EuropeanJournal of Sociology, 4 : 250–282. Reprinted in: The Sociology of Science: Theoretical and EmpiricalInvestigations. Chicago: University of Chicago Press, 1973, pp. 383–412.

Page 9: On the shoulders of Robert Merton: Towards a normative theory of citation

H. SMALL: On the shoulders of Robert Merton

Scientometrics 60 (2004) 79

MERTON, R. K. (1965), On the Shoulders of Giants: A Shandean Postscript. New York: The Free Press.MERTON, R. K. (1968), The Matthew effect in science. Science, 159 (3810) : 56–63. Reprinted in: The

Sociology of Science: Theoretical and Empirical Investigations. Chicago: University of Chicago Press,1973, pp. 438–459.

MERTON, R. K. (1969), Behavior patterns of scientists. American Scientist, 58 : 1–23. Reprinted in: TheSociology of Science: Theoretical and Empirical Investigations. Chicago: University of Chicago Press,1973, pp. 325–342.

MERTON, R. K. (1977), The Sociology of Science: An Episodic Memoir. Carbondale: Southern IllinoisUniversity Press.

MERTON, R. K. (1979), Foreword to GARFIELD, E., Citation Indexing – Its Theory and Application inScience, Technology and Humanities. New York: John Wiley.

MERTON, R. K. (1988), The Matthew effect in science, 2: cumulative advantage and the symbolism ofintellectual property. ISIS, 79 (299) : 606–623.

MORAVCSIK, M. J., MURUGESAN, P. (1975), Some results on the function and quality of citations. SocialStudies of Science, 5 : 86–92.

PALEVITZ, B. A. (1997), The ethics of citation: a matter of science’s family values. The Scientist, 11 (12) : 8.PRICE, D. J. D. (1965), Networks of scientific papers. Science, 149 (3683) : 510.SMALL, H., GRIFFITH, B. C. (1974), The structure of scientific literatures. 1: identifying and graphing

specialties. Science Studies, 4 (1) : 17–40.SMALL, H. (1978), Cited documents as concept symbols. Social Studies of Science, 8 (3) : 327–340.SMALL, H. (1984), Lives of a scientific paper. In: Selectivity in Information Systems. K. WARREN (Ed.),

New York: Praeger, pp. 83–97.SMALL, H. (2003), Paradigms, citations, and maps of science: a personal history. American Society for

Information Science and Technology, 54 (5) : 394–399.STENT, G. S. (1972), Prematurity and uniqueness in scientific discovery. Scientific American, 227 (6) : 84–93.