Does Science Push Technology? Patents Citing Scientific Literature

Abstract

In recent years, there has been a trend towards a science and technology policy that is applied in nature. In times of growing public discontent about high taxation and budget deficits, even basic science needs to document industry relevance. Counting patents apparently related to basic research activities through citations would be one way of measuring the relevance of basic research to industry. This has become especially interesting since Narin et al. [Narin, F., Hamilton, K.S., Olivastro, D., 1995. Linkage between agency supported research and patented industrial technology. Research Evaluation 5 (3), 183–187.] observed an increasing linkage between US technology and public science. The results of Narin et al. indicate a growing relationship between science and technology in a very general way. This idea of an increasingly science-based technology might convey the impression that there is direct knowledge-transfer taking place that is reflected in citations to scientific research papers in patents. A study of front pages of patents in the field of nanoscale technologies suggests that citation linkages hardly represent a direct link between cited paper and citing patent.

Full text

Ž.
Research Policy 29 2000 409–434 www.elsevier.nlrlocatereconbase
Does science push technology? Patents citing scientific literature
Martin Meyer ),1
SPRU, UniÕersity of Sussex, Mantell Building, Brighton BN1 9RF, UK
Received 30 April 1998; received in revised form 20 May 1999; accepted 20 May 1999
Abstract
In recent years, there has been a trend towards a science and technology policy that is applied in nature. In times of
growing public discontent about high taxation and budget deficits, even basic science needs to document industry relevance.
Counting patents apparently related to basic research activities through citations would be one way of measuring the
w
relevance of basic research to industry. This has become especially interesting since Narin et al. Narin, F., Hamilton, K.S.,
Olivastro, D., 1995. Linkage between agency supported research and patented industrial technology. Research Evaluation 5
Ž. x
3 , 183–187. observed an increasing linkage between US technology and public science. The results of Narin et al. indicate
a growing relationship between science and technology in a very general way. This idea of an increasingly science-based
technology might convey the impression that there is direct knowledge-transfer taking place that is reflected in citations to
scientific research papers in patents. A study of front pages of patents in the field of nanoscale technologies suggests that
citation linkages hardly represent a direct link between cited paper and citing patent. q2000 Elsevier Science B.V. All rights
reserved.
Keywords: Science; Technology; Citation; Patents; Nonpatent references
1. Introduction
The study of references in patent documents — a
form of linkage bibliometrics — has contributed to
the debate about science and technology and how
)Corresponding author. Fax: q49-231-593773; E-mail:
msmeyer@hotmail.com
1Also at the Institute of Strategy and International Business,
Department of Industrial Engineering and Management, Helsinki
University of Technology, P.O. Box 9500, SF-02015 HUT, Es-
poo, Finland, and Technopolis, 3 Pavilion Buildings, Brighton
BN1 1EE, UK.
they relate to each other. In this context, one has to
mention the pioneering work of Narin et al. They can
show an ever closer interaction between science and
Ž
technology in high-technology areas e.g., Narin and
.
Noma, 1985 . They also claim to have observed an
increasing linkage between US technology and pub-
Ž.
lic science. Narin et al. 1995 show that the within-
country connection between basic science and ap-
plied technology is especially pronounced in the
highly scientific areas of technology. A more recent
study by the same group underpins these findings
Ž.
Narin et al., 1997 . By tracing the rapidly growing
citation linkage between US patents and scientific
papers, Narin et al. reveal a strong national compo-
0048-7333r00r$ - see front matter q2000 Elsevier Science B.V. All rights reserved.
Ž.
PII: S0048- 7 3 33 99 00 0 4 0-2

()
M. MeyerrResearch Policy 29 2000 409–434410
nent, with each country’s inventors preferentially
citing papers authored in their own country by a
factor of between two and four.
How can one interpret these impressive citation
data? Is a citation linkage also a direct link? What
does it tell us about the nature of the interaction
between science and technology? Can this measure
indicate the direction of knowledge transfer? Does a
patent citation of a scientific research paper indicate
an immediate application of science? This paper
aims at contributing to a better understanding and
interpretation of the data generated by Narin et al.
Therefore, it investigates the histories of a number of
patented inventions and how the patent citations,
especially the science references, are related to the
invention process. Firstly, it will place citation analy-
sis in the general debate about the relationship be-
tween science and technology. After clarifying defi-
nitional issues, different interpretations of Narin’s
data will be introduced, one based on a basically
linear understanding of the innovation process, the
other one drawing on a more enlightened perception
of the relationship between science and technology.
The question will be raised as to whether one can
interpret citation linkages as a direct link indicating
science-dependence of technology. In the following
sections, the role of citations in patents will be
discussed in relation to the research issues of the
paper. Then a detailed discussion of the case selec-
tion follows. Based on findings from the cases and
additional interviews with patent experts, conclu-
sions will be drawn. One major point this paper will
make is that patent citations do not indicate a direct
science link of technology, but illustrate the multi-
faceted interplay between science and technology. A
further point is that one should demonstrate more
caution in interpretation even in technological areas
with high citation frequencies of scientific research
papers and not automatically assume that there is
only a linear one-way connection. One should not
use a misleading term like ‘science-dependence’ of
technology either. The results from the cases show
that there are several ways in which science and
technology relate to each other and that even in those
fields, technology is more than just a receiver and
transformer of scientific results. Therefore, it does
not seem appropriate to use the linear science-push
model to interpret patent citation data.
2. The relationship between science and technol-
ogy
2.1. Definitions
The question of whether articles cited on the front
page of a patent would be indicators of science bases
of the invention described by the patent is widely
addressed in the bibliometric literature. Rabeharisoa
Ž.
1992 notes the heterogeneity of the results of stud-
ies addressing this topic. Debates focus not only on
practical issues, but also on the very problem of
identifying and describing the relationships between
science and technology. To capture the relationship
between science and technology, citation studies usu-
ally establish a link through nonpatent literature cited
in patents. This means that patents are considered a
representation of technology, while papers and cita-
tions to them are viewed as representations of sci-
ence.
Traditionally, companies patent more than they
publish, and university researchers publish usually
Ž.
more than they patent. As Pavitt 1998a reports,
business firms are granted about 80% of all patents,
and many of the remaining 20% are granted to
individuals who are owners of small firms. Looking
Ž.
at publications, Hicks 1995 has established the
reverse pattern. Academics publish more than do
their colleagues in industry. In such a situation, with
most of the scientific papers published by academics
Ž.
and most patents held by and originated in indus-
trial companies, the tendency to associate the aca-
demic sector with science and the industrial sector
with technology is only natural. This could convey
the impression that science and technology are de-
fined organizationally. One might say that due to this
pattern, citation studies employ a quasi-organiza-
tional definition of science and technology.
2.2. Different interpretations
A debate on the relationship of science and tech-
nology without any reference to the linear model is
Ž.Ž .
hardly possible. Narin 1994 p. 150 acknowledges
that the old ‘linear model’ is ‘‘simplistic and highly
inaccurate,’’ and he also says that it ‘‘ignores the
much more intimate relationship between science
and technology’’.

()
M. MeyerrResearch Policy 29 2000 409–434 411
Then again, Narin et al. appear to have a basically
linear understanding of the innovation process. This
becomes apparent in the rhetoric they use to describe
the nature of the science–technology relationship
that is established by patent citations to scientific
research papers. For instance, they postulate ‘science
Ž
dependence’ of certain technologies Carpenter and
Narin, 1983; similarly, Grupp and Schmoch, 1992; p.
.
123 . Another problematic term is ‘scientification’ of
Ž.
technology see Grupp and Schmoch, 1992; p. 92ff .
In a way, Narin has pointed to this development
himself by indicating the ever closer relationship
between science and technology in another context.
In his common paper with Noma, he refers to Toyn-
bee’s dancer metaphor to describe the relationship
between science and technology:
... not only are technology and science intimately
related as a pair of dancers, but further that much
of current high-technology is becoming virtually
indistinguishable from science: Toynbee’s dancers
are today locked in an embrace from which it is
Ž
virtually impossible to separate the partners Narin
.
and Noma, 1985; p. 370 .
But with dancers dancing an ever closer dance, it
also gets increasingly difficult to say who is the
partner that determines the direction. This idea is
taken up in the two-branched model of innovation
Ž.
suggested by Rip 1992 . Its starting point is an
empirical or semi-empirical finding. Two different
Ž.
kinds of activities branch out from there: 1 ex-
ploitation understood as technological development,
Ž.
pilot processes and feedback, and 2 exploration to
increase understanding. The latter is done through
scientific research and can be considered a rational-
ization process: ‘‘The insights derived from the ex-
ploration branch may sometimes be called in to
assist and improve exploitation, and what can be
Ž
called ‘transformation of the exemplar’’’ ibid., p.
.
139 .
Fig. 1 illustrates the difference between the linear
interpretation and the two-branched approach. Basic
research, or science, produces a number of scientific
research papers, indicated here as copyright symbols
Ž.
q. Technology is materialized in patents, illus-
trated here through the symbol for registered trade-
Žw.Ž.
marks . Fig. 1 a describes one way of interpret-
ing the data of Narin et al. The increasing number of
scientific research papers cited in patents documents
the growing connection between the areas of science
on the one hand, and technology on the other — as
Ž.
indicated by the bold arrow. Fig. 1 b contrasts this
simplified interpretation by a more complex view of
the relationship between scientific papers and patents
following Rip’s model.
Based on the differences between these ideas, one
should ask whether it is possible to show a direct
link between certain patents and the scientific re-
search papers they quote. To what extent do they
Ž. Ž.
Fig. 1. A Linear interpretation B Two-way interpretation.

()
M. MeyerrResearch Policy 29 2000 409–434412
reflect actual input of basic research into a specific
invention? Can one draw direct connections between
Ž.
particular papers and particular patents? Fig. 1 b
raises such questions by indicating that there may, in
many instances, be an only and highly mediated
relationship between science and technology, even
though one might find occasionally direct relation-
ships between a patent and a paper it cites.
To answer these questions, one must have a closer
look at individual patents to make valid judgements
in this matter. In particular, one should investigate
why research papers are cited rather than counting
them under the assumption that every count makes
every penny spent on basic research more legitimate
and thus more relevant to industry. To do this, one
has to examine the patent-front pages in detail to
find out what the inventors cited, how and why.
3. The role of citations in patents
3.1. Patent documents
Before investigating citation studies in more de-
tail, a few introductory remarks on patents and patent
citations and their roles seem necessary. Ganguli and
Ž.
Blackman 1995 briefly review the structure of a
patent document, which consists of three elementary
Ž.
parts: 1 the title page with bibliographic informa-
Ž.
tion, 2 the text, which includes a description of the
invention, preferred examples in details as well as
Ž.
drawings, diagrams, and flow charts, and 3 the
claims. Given that I want to evaluate citation studies,
I follow their focus on the title, or front page. The
front page of a patent contains a set of bibliographic
data. 2For the citation studies in question, the bibli-
ographic information available on front pages of
patents is most important. There are two sets of
references to other patents and literature in patent
2Ž.Ž .
Walker 1995 p. 135 gives a detailed overview of what is
contained on the front page in a standard format: bibliographic
information, examination-process information, an abstract of the
Ž.
invention, and a drawing when applicable , the title, the name of
the inventor or inventors, the date of application, the application
number, a record of previous applications from which the applica-
Ž.
tion stemmed where appropriate , and the classification number
with any applicable cross-reference classification numbers.
documents — examiner references on the front page
and applicant references embodied in the full-text
Ž.
e.g., Narin and Noma, 1985 . Citation studies just
rely on the front page-examiner references since only
those are available in the databases the researchers
Ž.
used Schmoch, 1993 . Narin et al. also omit any
analysis of references in the body of the patents for a
theoretical reason: ‘‘Theoretically, the front page
references should be the most important ones on a
patent, since they are the ones relied upon.. .by the
Ž
examiner in establishing the patent’s novelty’’ Narin
.3
et al., 1997; p. 319 . However, as one patent
attorney points out, ‘‘it could be that connections to
a scientific base are more observable in the applicant
references, since these are in the disclosure and
might have been included for reasons other than
distinguishing the claims — for example, for the
Ž
purpose of attribution of earlier work’’ Gary
4.
Jordan, personal communication .
To better understand what Narin et al. really do
and what function these particular citations have, a
closer look at the role of citations in patents is
helpful. Narin et al. and other authors make use of
Ž.
certain, so-called nonpatent references NPRs on the
front pages of US patents. NPRs can encompass
references to a variety of nonpatent documents, such
as scientific papers, technical papers, conference pro-
ceedings, textbooks, disclosure bulletins, abstract
services, and so forth. Just a minority of patents
contains references citing nonpatent literature. A re-
cent study of the Norwegian knowledge base by
Ž.
Iversen 1998 , for instance, shows that only about
30% of Norwegian-originated US patents contained
NPRs. This is due to the fact that the stock of prior
art references of the Patent Offices predominantly
consists of patent literature. For example, the stock
of the German Patent Office has about 10% non-
Ž5
patent literature only Rainer Bertram, personal
.
communication . Consequently, the official searches
3Also Narin et al. assume great similarity between front page
and text science references. They base this assumption on a brief
Ž.
study by Narin and Olivastro 1988 which indicates that approxi-
mately half of the science references are on the front page.
4US patent attorney, now associate professor at the Department
of Business Studies, Uppsala University, Sweden.
5European patent attorney, Grunecker, Kinkeldey, Stockmaier
¨
and Schwanhausser, Munich, Germany.
¨

()
M. MeyerrResearch Policy 29 2000 409–434 413
of the examiners mainly reveal patent literature doc-
Ž.
6
uments Rainer Bertram, personal communication .
3.2. Patenting as social process
Ž.Ž .
As van den Belt 1989 p. 185 establishes, most
studies on the working of the patent system have
chosen ‘‘a naıve conventional view that the patent
¨
system does no more than bestow legal recognition
upon what already exists in a well-defined form’’.
Sociological studies of the patent system point out
that the organizational machinery of the patent sys-
tem itself ‘‘determines what is to count as a
patentable ‘invention’ and how large the extent of
Ž.
the protection should be’’ ibid. . In other words,
patenting is also a social process. There are a number
of different reasons that motivate patenting 7and a
number of different actors who exercise their influ-
ence on the shape of patents.
As it addresses a number of different readers with
different interests, the entire patent document can be
seen as ‘‘a compromise between different strategies’’
Ž.Ž.Ž.
Rip, 1986; p. 92 . Rip 1986 p. 91f distinguishes
Ž.
three different kinds of readers: 1 the officials of
Ž.
the Patent Office, who have to grant the patent; 2
the competitors in the field, who are eager for infor-
Ž.
mation about new products and processes; and 3
potential licensees, who must be interested in what
the patent offers. This readership structure has conse-
Ž
quences on the way patents are written and how
.
citations are selected . On the one hand, one has to
fence off, while on the other hand, one must indicate
Ž.Ž.
an interesting area. Rip 1986 p. 92 explains:
Although patentees may value the patent system
in general, in writing their own application they
6In practice, patent applications which have a closer link to
traditional fields of science, for example, chemistry, physics,
biology, are more frequently confronted with nonpatent literature.
7Ž.Ž .
Campbell 1989 p. 211f describes the variety of uses
Ž.
patents have and their effect on patenting: 1 some parties hold
Ž.
many small patents to cover their production; 2 other economic
Ž.
actors focus on key central patents; 3 others prefer not to patent,
but rather to manage their affairs by means of industrial secrecy;
Ž.
and 4 patents are also taken out as part of employee incentive
programs to promote inventiveness, resulting in a large number of
not necessarily workable minor parts.
will try to keep the box around their own inven-
tion as black as possible while still satisfying the
enablement requirement. Penetrability is not in
their interest. ... Perhaps it would be best to
characterise the rhetorical structure of patent ap-
plications as a fence of interests.
Ž.Ž .
Following Campbell 1989 p. 212 , one can
establish that ‘‘patents are clearly a strategic compo-
nent in the economic exploitation and development
of technology and an important feature of the social
8Ž.Ž
dynamics of technical invention’’. Rip 1986 p.
.
97 emphasizes the legal and thereby strategic char-
acter of patents: ‘‘Legal battles may now be fought
about interpretations of words or sentences’’. 9
3.3. Patents as legal documents: applicants’ and
examiners’ tasks
Patents are legal documents. Hence, patent cita-
tions do not work like citations in papers. Campbell
Ž.
and Nieves 1979 argue that, due to their specific
legal functions, citations in patents are likely to be
much more carefully selected than citations in jour-
nal papers. Due to the wide range of legal require-
ments, possible loopholes and eventual litigations,
‘‘patent applications are often written by specialist
patent advisers and not by the scientists themselves’’
Ž.
Rip, 1986; p. 91 . This section will discuss what the
different roles of patent applicant and attorney as
well as examiner are and how citations reflect them.
The legal responsibility of the applicant is to
Ž.
describe the prior art. As Collins and Wyatt 1988
Ž.
p. 66 explain, ‘‘he or she must set out the back-
ground in such a way as to show how the claimed
invention relates to, but is innovatively different
8This statement is well underlined by the difficulties Campbell
experienced when he tried to get cooperation from corporations
involved in patent actions. In particular, their objections included
the concern in jeopardizing their legal position in relation to the
Ž.
patent in question ibid., p. 234 .
9Ž.
For the field of biotechnology, Cambrosio et al. 1990
illustrate the strategic character of patents and, furthermore, show
that ‘‘while one might expect ‘technical’ arguments to play a
central role... ‘social’, ‘historical’ ‘economic’, or ‘philosophical’
arguments are coextensive with and constitutive of the ‘technical’’’
Ž.
ibid., p. 290 .

()
M. MeyerrResearch Policy 29 2000 409–434414
from, what was already public knowledge’’. Another
legal requirement for patents is their usefulness.
Thus, the applicant’s task is to identify work ‘‘either
related to, but significantly different from, or else a
useful step towards, the new invention or a use of the
Ž.
invention’’ Collins and Wyatt, ibid. . The third legal
requirement a patent has to meet is novelty.
The examiner’s task here is to ensure the novelty
of the invention claimed and to identify its limits.
Hence, the function of examiner citations is ‘‘to
identify the area in which the invention is truly
original and therefore merits the granting of a patent’’
Ž.
Collins and Wyatt, ibid. . Examiner citations usu-
ally complement, rather than duplicate the citations
by applicants, although ‘‘it happens not infrequently
that both examiner and applicant cite the same publi-
Ž.
cation’’ Collins and Wyatt, ibid. .
At this point, some criticism already emerges with
regard to the quality of citations. Despite the fact that
citations in patents are less likely to be irrelevant or
superfluous than references in journal papers, Collins
Ž.Ž .
and Wyatt 1988 p. 67 observe that they are not
free of problems either. ‘‘Examiners, for example,
tend to restrict their reading to a narrow range of
specialties and to be relatively unfamiliar with the
wider literature. Where they do use the journal litera-
Ž
ture it tends to be in the secondary form i.e.,
.
abstracting journals rather than the primary form.’’
These ‘bad’ citation manners of examiners 10 and
the resulting irregular nature of the front page NPRs
have made certain scientometricians — at least to
some extent — question the hypothesis that these
articles point to scientific bases of technology as
Ž
described in the respective patents van Vianen et al.,
.
1990; see also Rabeharisoa, 1992 . However, Narin
Ž.
et al. 1997 note the dramatic increase of NPRs in
10 Ž.
Collins and Wyatt 1988 also conclude from their observa-
tion of occasional repetitions of bibliographic errors that some-
times, examiners have not read the papers they cite. As they
sometimes found the use by examiners of the same set of citations
in several different patents, Collins and Wyatt also discern an
Ž.
‘‘occasional tendency to cite by rote rather by relevance’’ p. 67
and speculate that ‘‘some citations are included to demonstrate the
examiner’s diligence in searching rather than because of their
Ž.
direct relevance’’ p. 67 . National chauvinism in examiners’ and
applicants’ citation behavior is another observation by Collins and
Wyatt.
the course of the last few years. Already in the
Ž.Ž .
mid-eighties, Narin and Noma 1985 p. 373
pointed out that in the early 80s, the examiners had
been placing more and more references on the front
pages of the patents, and the applicants had been
required to be more meticulous in informing the
examiner about any references pertaining to the
patent.
3.4. Types of patent citations
Further criticism of citation studies is related to
the occurrence of different types of citations and
their frequency. One can distinguish a number of
different citations made in patent documents. As
Ž.
Schmoch 1993 explains, because of the novelty
requirement, the examiner has to look for earlier
Ž.
documents primarily patents that have the same or
almost the same features as the patent application.
Only if there are no other relevant documents putting
the novelty of the invention in doubt, will the patent
application be accepted. Hence, one can find two
Ž.
different types of citations: 1 documents of particu-
Ž.
lar relevance, and 2 references concerning the gen-
eral background. Documents of particular relevance
restrict the claims of inventors. In European search
reports, individual documents that, if taken alone,
may question novelty or inventiveness of a patent
claim are marked with the letter ‘X’. Documents
which are considered to question inventiveness of a
patent claim, if taken in combination with another
document, are marked with the letter ‘Y’. These
letters reflect the opinion of the search examiner, but
Ž
are not binding in the substantive examination Rainer
.
Bertram, personal communication . The second type
Ž
of references concerning the general, technical
.
background of the invention is marked ‘A’
Ž.
Schmoch, 1993 .
The different types of cited references have differ-
ent degrees of linkage, or proximity, to the examined
patents. While ‘X’ references have a high degree of
linkage, it may vary for ‘Y’ references, since they
are important only with other references. ‘A’ refer-
Ž.
ences usually have a low proximity Schmoch, 1993 .
As frequently cited, references are not necessarily
technically or economically important, but may be
cited for didactic and illustrative reasons in the de-

()
M. MeyerrResearch Policy 29 2000 409–434 415
Table 1
Ž.
Motives for citing nonpatent literature based on Grupp and Schmoch, 1992; p. 79f; and Schmoch, 1993
Motive Science link?
1. Prior art is not yet documented by a patent. Examiner thus relates progress in the examined patent Yes
application to a scientific publication.
Ž.
2. Citations of nonpatentable research results e.g., formulae, hypotheses, discoveries, etc. Maybe
3. Only nonpatent publications are available due to rapid development in a particular technical field. Maybe
4. Invention published in the company’s own journal or disclosure bulletin. No
Ž.
5. Citations of patent abstract services for Japanese language and chemical patents . No
6. Citations of very simple facts below the threshold of inventiveness so that no related patent exists. No
scription of prior art, further investigations about the
extent ‘X’, ‘Y’, and ‘A’ references occur are neces-
sary. A number of investigations have dealt with this
Ž.
issue Bare, 1981; Schmoch, 1993 . They show that
´Ž
relatively few citations less than 15% in both stud-
.
ies are references of the ‘X’ type. If one sees ‘X’
and ‘Y’ references as citations of a high–medium
proximity, 29% of all citations can be considered
Ž
‘‘linkages to direct knowledge sources’’ Schmoch,
.
1993 . So just about a third of all patent citations
have a close relationship to the citing patents. 11
Ž.
According to one survey by Schmoch 1993 ,
only 8% of all examiner citations originate from the
inventor. His results about citation types and their
frequency presented in Section 3.1 referred to refer-
ences citing to patents only. Although the categories
introduced there apply for nonpatent references
equally, for database reasons, it is not possible to
analyze NPRs automatically in the same manner.
This makes a closer look at the motives behind citing
to nonpatent references necessary.
3.5. MotiÕes behind NPRs
Ž
Interview-based surveys Grupp and Schmoch,
.
1992; Schmoch, 1993 could identify a number of
reasons for including nonpatent references in search
reports. Table 1 gives an overview of the motives
and evaluates to what extent they can be used to
make judgements on science link.
11 Ž.
However, Schmoch 1993 can present some evidence for a
more general relationship. He cites an intellectual evaluation study
of patent references that has shown for the field of space technol-
ogy that a large number of references are linked to the citing
Ž.
patents in a very broad sense Schmoch et al., 1991 .
Ž. Ž.
Grupp and Schmoch 1992 and Schmoch 1993
discuss to what extent the various motives are rele-
vant for a citation study that aims to indicate science
dependence of certain technological fields. Motive 6
is obviously not science-related. So is motive 5,
since it refers to patents just in form of abstracts.
Motive 4 refers to cases when the invention seems
not important enough to be patented, but competitors
should be prevented from applying for a patent on
the same subject. Therefore, publications in a com-
pany disclosure bulletin are similar to patents. The
respective citations then have quite the same charac-
ter as patent references. Motives 2 and 3 may, but do
not need to, relate to a science-intensive field, while
Ž
motive 1 ‘‘reflects the classic situation’’ Grupp and
.
Schmoch, 1992; p. 80 . However, even here, Grupp
Ž.
and Schmoch 1992 warn:
... care must be taken to ensure that not every
journal is tantamount to a reference to fundamen-
Ž
tal research. According to a convention Marquis
.Ž .
and Allen, 1966 p. 80 , a journal is only re-
garded as ‘scientific’ if it is published by a scien-
tific society. Journals from technical socie-
ties. ..on the other hand reflect the area of ap-
plied research.
In the case of technical journals, Grupp and
Ž.
Schmoch 1992 refer to Gibbons and Johnston
Ž.Ž .
1974 p. 228 who assume that technical literature
may also contain basic research reports. In light of
this discussion and no quantitative data on the impor-
Ž.
tance of the various motives, Schmoch 1993 con-
cludes that the quota of nonpatent citations as an
indicator for the science–technology link has to be
interpreted very carefully.

()
M. MeyerrResearch Policy 29 2000 409–434416
3.6. Research issues of this study
Not all of these criticisms affect the latest work
by Narin et al. Although their work is based on an
analysis of nonpatent references, which were listed
as ‘other references cited’ on the front pages of US
patents, they do not use the entire set of references
Ž.
Narin et al., 1997; p. 318 . Narin et al. standardized
the NPR data in several steps. 12 To a certain extent,
this procedure took care of the irrelevant motives
listed by Grupp and Schmoch.
However, as the discussion of previous work has
shown, the kind of link a patent citation of a scien-
tific research paper establishes is of a highly medi-
ated nature. Authors use the term ‘science link’ in
terms of a prior art andror claim-restricting relation-
ship. However, if proponents of this method use
terms such as ‘science-dependence’ or ‘science-base’
of technology, their readers could form an overly
simplistic impression of this mediated relationship. 13
Ž.Ž .
Narin et al. 1997 p. 317 refer to public science as
a ‘driving force’ that stands behind high technology
and economic growth. They claim to provide quanti-
tative evidence for the direction of that force. But is
it possible to establish such a relationship if one
12 First, the data were categorized into different sets, such as
science references, abstracts, and books. They identified roughly
242,000 out of 430,226 NPRs as science references defined as
‘‘citations to scientific journal papers, scientific meetings and
Ž.
other scientific publications’’ Narin et al., 1997; p. 319 . Then,
Narin et al. formed another subset of the science references,
containing citations to papers published in the journals of the
Ž.
Science Citation Index SCI . The 175,000 SCI science references
underwent further standardization and finally, authors’ addresses
belonging to the remaining NPRs were linked up with public
sources of financial support.
13 Direct knowledge transfer is meant here as a direct link.
Using the terms knowledge or science link, not to mention
science dependence, might confuse even a specialist audience in a
field related to ‘linkage bibliometrics’. When an earlier version of
this paper was presented to an audience of bibliometricians, a
number of them were surprised, if not ‘shocked’, to learn that the
Ž.
citation measure discussed was based on examiner not applicant
references because they — exposed to linkage rhetoric — auto-
matically assumed a direct knowledge link between the invention
and the paper cited. As indicated in Section 3.1, applicant refer-
ences in the disclosure might be more likely to contain attribution
of earlier work.
looks at individual inventions? Does science push
technology?
Ž.Ž .
As Narin et al. 1997 p. 319f point out them-
selves, their data measure linkages in codified
knowledge only. Several writers in the literature of
innovation and technical change have emphasized
the importance of tacit components of technological
knowledge and that information contained in scien-
tific papers often will not suffice to implement the
Ž
technology in question David, 1993; Howells, 1996;
see also, e.g., Pavitt, 1987; Rosenberg, 1990; Arora,
.
1991 . This suggests that there are more types of
linkages between science and technology. One could
expect to find a multifaceted interplay between sci-
ence and technology in concrete situations. If this is
indeed the case, one must think about what conclu-
sions one can draw from NPR citation data with
respect to the direction of the knowledge flow.
One has to interpret Narin’s impressive data in a
different manner given the complexity of the sci-
ence–technology relationship. To what extent do
citation linkages reflect this complex interplay? Could
one still speak of the ‘old pusher science’ and sci-
ence as ‘force’, or would less propulsive metaphors
be more appropriate?
4. Method
4.1. Case selection
Ž.
According to Yin 1994 , case studies are particu-
larly useful when complex processes are to be under-
stood. As this study attempts to investigate the rea-
sons for citing scientific literature and shed some
light on patent practices, the case study format seems
to be most appropriate. However, a study that re-
quires interviews with actors by its nature restricts
the scope and scale of an investigation. In this
situation, even more attention must be paid to the
selection of the patents to be examined. Assuming
that there is a higher probability to find direct knowl-
edge links in a young and strongly science-related
area, this study of 10 cases investigates citations on
data sheets of US patents in the emerging field of
nanotechnology. Citation studies dealing with the
relationship between science and technology often

()
M. MeyerrResearch Policy 29 2000 409–434 417
Ž
focus on science-based technologies e.g., Narin and
Noma, 1985; Collins and Wyatt, 1988; Grupp and
.
Schmoch, 1992; Noyons et al., 1994 . Such a focus
is a bias if one wants to say something about patents
in general, but a useful entrance point to investigate
the phenomenon of science-dependence or science
Ž.Ž.
links. Collins and Wyatt 1988 p. 68 , for instance,
point out that they restricted their research on cita-
tions in patents to a field that had to be ‘‘young,
rapidly growing and strongly scientific’’.
As this study is of an exploratory nature, it em-
ploys broad selection criteria. In selecting the cases,
Collins’ and Wyatt’s criteria provided a guideline for
this study. The 10 cases are taken from the field of
Ž.
nanoscale technologies. Braun et al. 1997 describe
nanoscience and nanotechnology in their bibliomet-
ric study as a young and emerging area with hardly
any publishing activity in the 1980s and an exponen-
tial growth pattern observed in the 1990s. In a later
study, Olle Persson and I confirmed the results by
Braun et al. and found some evidence for the sci-
Ž
ence-related character of the field Meyer and Pers-
.
son, 1998 . More than 60% of the nanopapers we
found belonged to the category of natural sciences.
Just around 20% of the papers belonged to an ex-
plicit engineering discipline. 14 In the same study,
we were able to identify more than 2000 patents with
search terms based on our bibliometric study. If there
is any direct link to be shown, one should expect to
find it in a field like this, with a total production of
around 5000 papers and 2000 patents.
Nanotechnology was also chosen because it con-
stitutes more a cluster of technologies than a single
Ž.
technology e.g., Budworth, 1996 . The term ‘nano-
technology’ encompasses leading-edge developments
in a number of generic technologies that are novel in
nature because of the changing conditions and prop-
erties of structures at the nanometer scale. By choos-
ing patents from this field, one can compile a selec-
14 In this study, we used data from the Science Citation Index,
whose science bias is generally acknowledged. However, our
bibliometric results are in line with qualitative data in other
Ž
studies e.g., Kuusi, 1994; Bachmann, 1995; Budworth, 1996;
.
Tolles, 1996; Anttinen and Lounasmaa, 1997; Malsch, 1997
pointing out the importance of scientific developments in this
field.
tion of patents in a dynamic, novel, and strongly
science-related area. Some of the patents belong to
the field of nanostructured materials, others are re-
lated to opto-electronics and medicine. There are
also cases at the borderline with other fields, such as
biotechnology or microelectronics.
Previous studies discussed differences in citation
patterns between the US and other countries. There
Ž
was a lively discussion in the literature e.g., Grupp
.
and Schmoch, 1992 as to whether the differences
observed were due to actual differences between the
countries or due to language and other biases. Some
studies by Narin et al. also pointed out national
differences in science dependence within a defined
subarea of technology. This is unlikely, and their
results can be explained by the fact that the papers
cited relate to US data. 15 Findings from interviews
with patent experts, which are presented in Section
5, will shed more light on this issue.
This study is meant as an entrance point to study-
ing different patent practices. The debate about
whether or not the different citation patterns ob-
served are due to differences in national innovation
systems also relates to the question of different
patent practices in small economies. Surprisingly,
small open economies are not reflected in the studies
discussed earlier. Thus, it is interesting to have a few
US patents of Swedish inventors in addition to Ger-
man-originated US patents. Section 5 will point to
eventual differences between German and Swedish
examining and patenting practices.
15 Ž.
van Vianen et al. 1990 showed a clear English language bias
causing difficulties in comparing citation data between English
Ž.Ž
and non-English speaking countries. Collins and Wyatt 1988 p.
.
67 speak of ‘national chauvinism’ in applicants’ as well as
examiners’ citation behavior. As for citations to company jour-
Ž.
nals, Grupp and Schmoch 1992 point out that ‘‘similar refer-
ences to European or Japanese journals are unlikely to be cited
because Americans are probably not able to understand the respec-
tive languages of the country and the sources required are not
available in their card index. This argument too confirms that the
above-mentioned independence of the science connection from
national technology also applies there despite the contradicting
Ž.
results published’’ p. 92 . For a discussion of the effect of
different patent law formalities on patent citations, see Grupp and
Ž.Ž .
Schmoch 1992 p. 92f . See also Section 5.3 for a brief discus-
sion.

()
M. MeyerrResearch Policy 29 2000 409–434418
Table 2
Patent cases selected
Case US patent no. Invention patented Assignee Inventor
a
1 5,566,197 Tunable gain coupled laser device Multinational company UniversityrPSR
2 5,367,274 Quantum wave guiding switch Multinational company UniversityrPSR
3 5,418,197 SiC whisker and particle reinforced ceramic Large firm UniversityrPSR
and cutting tool material and whisker and particle
5,420,083 reinforced ceramic cutting tool material
4 5,603,958 Pharmaceutical carriers Foreign company UniversityrPSR
5 5,427,767 Nanocrystalline magnetic iron oxide University-affiliated UniversityrPSR
particles-method for preparation and use in institute owned by
medical diagnostics and therapy industrial company
6 5,559,353 Integrated circuit structure having at least one Multinational company Industrial researchers,
CMOS-NAND gate and method for the in-house
manufacture thereof
7 5,298,760 Performance of location-selective catalytic Multinational company Industrial researchers
reactions with or on the surfaces of solids in in-house
the nanometer or subnanometer range
8 5,543,289 Methods and materials for improved high Inventor who exploits Industrial researchers,
Ž.
gradient magnetic separation of biological patent in his firm SME in-house
materials
Ž.
9 5,470,910 Composite materials containing nanoscalar i PSR; PSR researchers,
Ž.
and particles, process for producing and their use ii specialist affiliate of in-house
5,590,387i for optical components and method for multinational company
producing metal and ceramic sintered bodies
and coatings
10 5,250,207 Magnetic ink concentrate Multinational company Industrial researchers,
and in-house
5,500,141
aPSRspublic sector research organization.
Ž.
Grupp and Schmoch 1992 , in accordance with
Ž.
Dosi 1988 , believe that the various technological
fields have a much greater impact on patenting prac-
tices. Differences in citation frequencies between
technological fields would not be very surprising. US
Ž
and German patent studies see e.g., Meyer-Krahmer
.
and Schmoch, 1997 observed that pharmaceutical
patents cite scientific papers much more than me-
chanical and automobile patents. This again is in line
with the idea of different patterns of innovation as
suggested by Pavitt in his technology-based classifi-
Ž
cation of business firms Pavitt, 1984; Bell and
.
Pavitt, 1993 .
One can even find indications for the notion of
different, technology-specific patenting practices
within companies. Preliminary research indicates that
different patent practices within a firm depends on
the division of the company that is applying for a
patent. A researcher of a large German multi-tech-
nology firm, for instance, describes the patenting
behavior of his colleagues in mechanical engineering
in this manner who tend to patent processes: ‘‘They
apply for a patent when they have found a new way
Ž16
of driving a screw into something’’ N.N., per-
.
sonal communication . This view makes it interest-
ing to have a look at patents that originated in the
same company, but in different technological fields.
In broad areas, such as materials, there might be
differences within the company even between sub-
fields. This survey includes two patents that origi-
nated in the same company, but have applications in
Ž.
two different areas cases 7 and 10 .
The cases reflect the variety of inventors and
assignees and the different institutional affiliations of
Ž.
the inventors. Individual assignee–inventors case 9 ,
16 Research manager in a large German multitechnology com-
pany.

()
M. MeyerrResearch Policy 29 2000 409–434 419
Ž. Ž .
weaker case 4 or stronger ties cases 1–2, 5 of
external inventors to their assignees are also in-
cluded. Table 2 gives an overview. Different patent
practices might also exist between inventors who
have a background in academic research and indus-
trial researchers. There might be a tendency to in-
clude more science references in the specifications
with academic researchers than with industrial inven-
tors. This aspect has also played a role in the case
selection.
Ž.Ž .
Rabeharisoa 1992 p. 68 remarks that the exis-
Ž.
tence or nonexistence of patents in a field is not a
trivial result since ‘‘the existence or nonexistence of
patents indicates a certain structuration of the field
studied’’. Following this idea and applying it to
Ž
patent citations, this would mean that existence or
.
nonexistence of NPR or science references is not a
trivial finding, either. Thus, it is useful to include
also patents that — although being in a supposedly
science-based field — do not cite to scientific litera-
ture, but exclusively refer to other patents.
The selection of patents also reflects the variety of
inventors and assignees. The origin of the respective
inventors could play a role in their citation behavior,
which might be reflected by the examiner references
at least to a certain extent. University researchers
might refer more frequently to papers than their
counterparts in R&D labs of companies — who
might be more drawn to patents, for instance. Thus,
the sample includes inventors who are academics as
well as industrial researchers. On the assignee side,
large firms, small and medium-sized companies, as
well as research institutes and individuals are in-
cluded. The focus in this purposive sample is on
large firms, since they not only carry out most of the
R&D activities, perform and publish most of the
basic research based in industry, but also maintain
Ž
the closest links with academic research Hicks,
.
1995; Pavitt, 1998a . Interviews with patent experts
suggest that the existence of different organizational
structures, e.g., the patent department, might have
some impact as well. While independent inventors
who are not employed by a corporation appear to be
mainly interested in ‘‘showing that their invention is
better than the prior art and is so clever that a patent
should be granted,’’ the corporate inventors, who
have this same motivation, also have been trained
over the years by their attorneys to mention the
Ž
closest prior art in their disclosure documents Gary
.
Jordan, personal communication .
Table 3
Nonpatent references and citations to patents in cases selected
Case Nonpatent references Citations
to patents
a
Journal Books Company Conference Abstract Others Total Total
b
articles journals proceedings services
112–1–165
28112––1211
3 5,418,197 1 – – – – – 1 7
5,420,083 2 – – – – – 2 8
42–––––25
5––––––025
62––––136
73–1–––41
85–––––525
9 5,470,910 – – – – – – 0 8
5,590,387 – – – – 1 2 3 13
c
10 5,250,207 – – – – 3 – 3 8
5,500,141 – – – – – – 0 13
Average 1.9 0.2 0.2 0.2 0.3 0.3 3.2 10.4
aIncluding advanced level textbooks.
bCompany disclosure bulletin.
cTwo citations to abstract services for Japanese language patents, one citation to an abstract service for chemical patents.

()
M. MeyerrResearch Policy 29 2000 409–434420
Table 4
Findings from the cases: science links
Ž.
Science links Indicated by NPRs? Mentions reported in cases. ..
Ž.
ØInventions based on general experience in research and teaching No 5 1–2, 4–6
Ž.
ØScientific literature important as underlying information to inventors No 3 3, 5, 7
Ž.
ØInventions stimulated by articles in downstream specialist journals No 1 2
Ž.
ØResearch papers as background information cited in patents Yes 1 7
Ž.
ØScience references used to attack, restrict, or modify claims Yes 1 1
Ž.
ØResearch papers cited in special cases only Yes 2 4, 9
4.2. Data collection
The focus of this work is on whether there is a
direct link between a patent and scientific research
papers it cites, which supposedly embody the scien-
tific research relevant to the invention patented. One
can only investigate this relationship if one talks to
the inventors. Any links established by examiners or
patent attorneys are not directly related to the inven-
tion process and thereby have to be viewed as a
mediated connection between patents and underlying
science.
The inventors were contacted to give information
on the invention process, the resulting patent, and the
connections their invention has to science. Specifi-
cally, they were asked to describe the background of
the invention, relate the citations in the patent to
Ž.
their invention if possible , and indicate what im-
pact those references had. Furthermore, it was in-
quired to what extent the patent citations refer to
knowledge sources that had played a role in the
invention process and what other factors not indi-
cated by the references had influenced the invention.
The inventors were contacted in a number of
ways. Information on the motivational factors of
citation behavior as well as on the background of the
invention itself was gathered through semi-structured
interviews carried out in person, by telephone and
e-mail.
Table 3 summarizes the citation characteristics of
the patents selected. The patents included in this
survey on average contain two times more nonpatent
references than the average patent in Narin’s sample
Ž.
3.2 NPRs per patent vs. 1.5 NPRs per patent . One
could classify this selection of patents as ‘science-
based’ in Narin’s terminology. Considering the cited
journal articles and conference meetings as well as
the books in the patents of this survey as science
references, the average rate is 2.3 science references
per patent. 17 The patent cases in this survey have
about the same amount of references citing to other
Ž
patents as does the average patent described in Narin
.
et al., 1997 : 10.4 referencesrpatent vs. 10.5 refer-
encesrpatent.
5. The cases
5.1. Science links
The data from the cases show that there are
indeed many links between science and patents. Table
4 lists the observations from the cases with regard to
science links. It specifies for each science link
whether it is indicated by an NPR and reports how
many times this particular link has been mentioned.
The numbers in parentheses refer to the individual
cases, which are summarized in Appendix A. One
proposition derived from the literature review was
that patent citations of research papers usually estab-
lish only a highly mediated link between science and
technology. The cases support this idea, even though,
in many instances, inventors did say that science and
scientific research papers had played a significant
role with regard to their inventions. In five of 10
17 This is much higher than the average NPRrpatent rates given
Ž.
in Narin et al., 1997 . It is still higher than most of the 1995
average science references per patent for particular technology
Ž.
fields as reflected by German patents Narin et al., 1997; p. 321 .
This survey includes also three out of 13 patents that do not
contain any NPRs; these cases, of course, lower the NPRrpatent
rate and the science referencerpatent relation.

()
M. MeyerrResearch Policy 29 2000 409–434 421
cases, inventors indicated the general importance of
science. They stated explicitly that their inventions
are based on their general experience in research
andror teaching. The three inventors in cases 3, 5,
and 7 characterized scientific research papers as
important, but background information only.
Only in one of the 10 cases that one can draw an
‘antecedent’ cognitive link between a patent and a
particular publication that has stimulated the inven-
tion patented. In case 2, the inventor hardly recog-
nized some of the examiner citations, but was able to
remember an article he read in Physics Today which
initiated work on the invention patented. Case 2 is a
demonstration of knowledge transfer between two
technological areas that was stimulated by more
downstream research news. 18 The paper was not
cited on the front page, since it was not relevant to
the examiner in terms of the three legal requirements
for granting a patent.
While patents actually contain NPRs, they do not
reflect the cognitive contribution of a particular pa-
per to the invention. As for reasons behind citing
nonpatent literature, the interview data of this study
do not contradict the motives given by Grupp and
Ž. Ž.
Schmoch 1992 and Schmoch 1993 . Most of the
patent citations in this study can be categorized
accordingly.
The most direct connection between patent and a
cited paper is to be found in case 7. Here, a patent
that resulted from a basic research project in which a
few papers were also written, one of which happened
to be cited by the examiner as background informa-
tion. The paper itself is another result of the inven-
tor’s work in that particular area that also led to a
patent rather than that showing a direct contribution
of science to technology. Another science-link is
established by the examiner who finds references in
Ž
the scientific literature that restrict a claim e.g., case
.
1 . The case illustrates the patent examiner’s role as
‘‘the primary source of references in that he or she
develops them in course of searching for prior art’’
Ž.
Campbell and Nieves, 1979; p. 963 . Here, the
18 This supports Whitley’s view that ‘popularized’, more down-
stream knowledge feeds back into research and scientists learn
about fields outside their immediate research areas from more
Ž.
popular accounts Whitley, 1985; see Hilgartner, 1990 . This, too,
is an example for mediated knowledge transfer.
inventor had to modify his claims after the patent
examiner had confronted him with material restrict-
ing parts of the original claims. Two other inventors
Ž
said they used citations of papers in special cases 4,
.
9 . Here, using NPRs can be seen in context of
Ž.
strategic thinking ‘fencing off’ and the legal char-
acter of patents.
Other cases substantiate the idea of organizational
influence on citations and thereby underline the no-
tion of patent citations of research papers as a highly
mediated linkage between science and technology. In
the present cases, these influences are either the
Ž.
patent departments of large firms cases 6, 7 or the
specialized patent attorneys on behalf of high-tech-
nology SMEs who make suggestions for adding ref-
erences, and thus making the patent ‘water-proof’
Ž.
case 8 . These findings clearly confirm the idea that
there is no direct, ‘antecedent’ relationship between
cited paper and citing patent. In the cases, the scien-
tific literature cited rarely seems to be the original
source of the idea that brought about the invention.
The science link established by patent citations is
indirect in the sense that the link is not connected to
the invention process and the origin of whatever is
patented. A citation link connects patent and cited
paper insofar as it tells whether the invention touches
an area where there has been no patent before, but a
scientist has published something relevant.
5.2. Direction of the knowledge flow
The second issue this paper deals with is the
direction of the knowledge flow between science and
technology. Does science really push technology?
And to what extent do patent citations reflect this
science-push idea?
The review of previous work in the patent cita-
tions literature has shown that citation links between
patents and the papers they cite are, if not explicitly,
at least implicitly viewed as an indication of the
contributions of science to technology. Here, the
findings from the cases are not in complete accor-
dance with prior work. Table 5 summarizes the
observations from the cases. Inspired by Rabeha-
risoa’s work in a different theoretical context, an
attempt has been made to investigate whether it is at
all possible to differentiate scientific from technolog-
ical activities and actors on an individual as well as
an organizational level.

()
M. MeyerrResearch Policy 29 2000 409–434422
Table 5
Findings from the cases: examples of science–technology interactions
Types of science–technology interaction as reported in cases Reported in cases. ..
1. Close personal science–technology linkage: individuals active both in academic research and 1–2
Ž.
industry adjunct professorships of industrial researchers .
2. Doctoral candidate working in university-based corporate research institute. His work on 5
Ž.
the same subject-matter has led to both scientific results his PhD dissertation and
Ž.
technological output the patent examined .
3. Public research institute is actively involved in both patenting and publishing. Not 9
infrequently, patents and papers result from the same project.
4. Invention patented was developed by a scientist who was active in a company and produced 7
both patents as well as scientific research papers based on the same project.
The cases were taken from a field where high
Ž.
science-‘dependence’ in Narin’s sense could be
expected. In our view, the findings do not contest the
strong science-relation of the patented technology in
question. But they do question the assumed direction
of the knowledge-flow from science to technology
or, from academe to industry. As observations 1 and
2 as well as 3 and 4 indicate, researchers seem to
integrate scientific and technological activities in-
creasingly by working on one subject-matter, but
generating scientific papers as well as technological
Ž.
outputs patents . In such a situation, one might find
it difficult to judge whether science pushed technol-
ogy or technology pulled science. It seems to be a
much more reciprocal relationship than the linear
model suggests. In these cases, patent citations re-
flect much more than scientific contributions to tech-
nology. They also indicate the kind of closeness
between the two spheres. For instance, they illustrate
a close personal science–technology linkage by indi-
viduals working on one subject-matter in both scien-
tific and industrial organizations, as shown in obser-
Ž.
vation 1 cases 1 and 2 . They also point at simulta-
neous scientific and technological activities of indi-
Ž.
viduals in hybrid observation 2 , or established or-
Ž.
ganizations observations 3 and 4 . In these cases,
science and technology seem to be much more inter-
twined. This interwovenness is much better reflected
by Rip’s two-branched model of innovation with its
emphasis on interconnected processes of exploitation
and exploration than limited linear approaches.
In addition, not all the organizations’ researchers
are working for easily fit into a simple academerin-
dustry dichotomy anymore. Are a university-based
Ž.
corporate research institute observation 2 and a
Ž.
public research institute observation 3 , both of
which generate scientific papers as well as patents,
considered to be locations of scientific research or
technological development? Here is where one sees
the limitations of a quasi-organizational set of defini-
tions, and the potential usefulness of a cognitive
approach.
It was investigated how the inventors viewed the
relationship between science and technology in their
field of expertise. It is not possible to establish a
general tendency in favor of a science-push idea.
Quite the contrary seems to be the case. In two
cases, the inventors tended to speak rather of a
technology-driven science than a science-based tech-
nology by pointing out that ‘people patent before
Ž.
they publish’ case 8 or ‘patents in very advanced
and modern fields usually do not contain references
to scientific research papers’ and it is not technology,
Ž.
but science that is lagging behind case 10 .
In addition, in the citations of this survey, there
seem to be more differences between various techno-
logical fields than between the different assignee–in-
ventor combinations. The electronics-related patents
have a higher level of nonpatent references than do
the materials-related patents investigated. Different
technological fields have different citation patterns.
A good illustration of this idea is given in cases 7
and 10. Both cases include patents that originated in
the same company. Both inventions were developed
at the same site, but in different technological areas.
While the first case relies also on NPRs, the other
case exclusively reports on citations of other patent
documents. If one looks not only at the ‘other refer-

()
M. MeyerrResearch Policy 29 2000 409–434 423
ences’, but also at the citations to other patents, one
finds at least one strong variation in the citation rate.
Ž.
The biotech-related patent case 8 was a clear out-
lier with 25 references to other patents.
5.3. National differences in examining and patenting
practices
‘Linkage bibliometricians’ look at statistical ag-
gregates, not at single cases. If one wants to make
the case that there are no direct links between cited
publications and citing patent, one has to explain
how other prior work can establish a within-country
Ž
connection between science and technology Narin et
.
al., 1995, 1997 , and in particular, one has to address
how other authors could observe local proximity
Ž.
19
between citing and cited patents Jaffe et al., 1993 .
This section will address the within-country link at
first and then deal with the local proximity observa-
tion.
As for the within-country connections between
patents and the scientific articles they cite, an inter-
esting observation is the strong variations in citation
frequencies between highly industrialized countries
Ž
in the same technological field e.g., see Narin,
.
1994 . If there was a direct link, one could assume
that a certain technology is more dependent on sci-
ence in one country than in another. The idea of a
highly mediated linkage would suggest that there are
other explanations for the observed differences in
citation frequencies.
This study encompasses European-originated US
patents only. However, different countries have dif-
ferent patent systems. 20 National patent practices
matter even if one looks at US patents only, since
most attorneys prepare a first application in the home
19 I would like to thank one of my referees for this annotation. It
has initiated another set of interviews, which has just further
confirmed my strong opinion about the mediated nature of citation
linkages.
20 This section chiefly reports on differences between US and
European patent practices, since this study deals with US patents
of European inventors. The situation in Japan is so different. A
corporate researcher there is required to deliver one to two
inventions per month. To meet these targets, Japanese inventors
have to submit patent applications that would not qualify for
granting patents elsewhere. According to a German patent exam-
iner, ‘‘60% of the new patent applications would not be granted in
Ž.
Germany’’ Hans-Hermann Zitt, personal communication .
country and then file that same disclosure in multiple
countries within 12 months under the Paris Conven-
Ž
tion on patents priority Gary Jordan, personal com-
.
munication . A number of interviews with patent
experts suggest that nationally different patenting
practices might affect citation frequencies measured
Ž21 22
Kristiina Gronlund, Gary Jordan, Leif Karlsson,
¨23 .
Hans-Hermann Zitt, personal communications .
Higher citation frequencies in US-originated
patents as compared to European-originated ones
might be a consequence of the way how patent
applications are examined generally. In Germany, for
Ž
instance, at first the main claim is examined Rainer
Bertram and Hans-Hermann Zitt, personal communi-
.
cations . The dependent claims are rather briefly
examined, mostly with one or two prior art docu-
ments disclosing the respective features. Many times,
the dependent claims are objected to without a de-
tailed reasoning why these claims were patentable or
Ž.
not Rainer Bertram, personal communication .
Regarding the examination at the European Patent
Office, the main claim and the dependent claims are
examined, but ‘‘we try to have one single claim that
Ž
functions as an umbrella’’ Leif Karlsson, personal
.
communication . The examining practice is rather
similar to the German one. Most important is that the
main claim meets the three criteria for granting a
patent — novelty, nonobviousness, and industrial
Ž
applicability Rainer Bertram, personal communica-
.
tion .
By contrast, in the US, all claims — including
Ž
dependent claims — are examined Leif Karlsson
.
and Hans-Hermann Zitt, personal communications .
The US examiners are obliged to examine each
claim thoroughly and to provide a detailed reasoning
about patentability of each claim in the first Official
Ž.
Action Rainer Bertram, personal communication .
However, the most important reason for different
citation frequencies in US-originated and European-
originated patents is the so-called ‘duty of disclo-
sure’ in the US. In Europe, it is up to the applicant to
21 Head of Consulting, PRH National Board of Patents and
Registration, Helsinki, Finland.
22 Patent attorney, Groth and Company, Stockholm, Sweden.
23 Deutsches Patemtamt — German Patent Office, Munich,
Germany.

()
M. MeyerrResearch Policy 29 2000 409–434424
introduce prior art known to him in the examination
Ž
procedure or to refrain from doing so Kristiina
.
Gronlund, personal communication . By contrast, US
¨
law stipulates that the applicant has to cite any prior
documents known to him to the USPTO as long as
the application is under examination. Noncompliance
with this requirement is considered as fraud by the
USPTO and can be used as a grounds for invalidat-
Ž
ing the patent Rainer Bertram, personal communica-
.
tion . To practitioners, ‘‘it is only logical that this
leads to different practices how patent attorneys pre-
Ž
pare the applications’’ Rainer Bertram, personal
.
communication . Therefore, this difference can be
seen as one chief reason why citation frequencies in
US-originated patents are higher than in European-
Ž
originated ones Leif Karlsson, personal communica-
.
tion . A good illustration of this observation is case
8, an invention that was German-originated, but
whose patent application was handled by US-based
specialist patent attorneys from the beginning. With
Ž.
25 references to other patents plus five NPRs , it is
the clear outlier in this survey. 24 One US patent
attorney reports:
Some years ago, we used to talk about ‘continen-
tal-practice’ type disclosures when dealing with
Ž
European origin patents. Specifications the dis-
.
closure portion of the patents written in the
‘continental’ style were much shorter than those
in the strictly U.S. style. The European patent
attorneys seemed to be writing the specifications
in a very pointed manner to a single new technical
point. Most of these specifications did not have
citations of literature at all, but when that was
done it was only one or two references. The U.S.
practice has been to write much more detailed
specifications and this usually meant commenting
Ž
on 5 or 6 pieces of prior art Gary Jordan, per-
.
sonal communication .
Drawing a comparison between US and continen-
Ž.
tal especially German patent practices, one of the
patent experts states:
the U.S. was regarded as a much tougher legal
environment. There had to be as much back-
24 This patent happens to be the only one in this study that was
processed by US specialists from the beginning.
ground information as reasonably possible to con-
vince the patent examiner that the prior art was
studied closely before filing the application, there
were very good distinctions between the claims
drawn and the prior art disclosures, and in the
event of future litigation there would be good,
meaningful distinctions that could be relied upon
Ž
in a legal battle Gary Jordan, personal communi-
.
cation .
For patent-to-paper citations, a localized pattern
of citations to papers probably means that problem
solvers ‘enter’ world science through local scientists.
Ž.
Using patent citations, Schild no date , however,
cannot observe such a pattern in her study of inven-
tiveness in the Swedish region of East Gothia. 25
This leads to the question of how to interpret highly
aggregated statistical regularities with regard to sin-
gle regions. The most likely explanation might be
found in national examining practice. 26
Swedish patent examiners prefer to cite a corre-
sponding English-language patent to a Swedish or
other-language patent because of a potential later
disclosure in multiple countries. Anticipating that a
Swedish applicant might wish to file the Swedish
25 Inspired by the work of Jaffe et al., Schild used a similar
method to study the impact an individual university has on the
region. The idea was ‘‘to identify the degree to which Linkoping
¨
Ž
University research was cited in East Gothia’’ Schild, no date; p.
.
38 . For the 1996 set of East Gothia patents, she identified 106
Ž.
citations including NPRs and 163 cited patent inventors. But
only seven of the 163 cited inventors had addresses in Sweden,
and of these, only one had an East Gothia address. Most of the
citations were to US inventors, with Japan and Germany trailing
behind the US.
26 Another possible explanation could be based on data pre-
sented in the Second European Report on Science and Technology
Ž.Ž
Indicators European Commission, 1997 Chap. 7, in particular,
.
p. 359ff . The report suggests distinct differences in science,
technology, and innovation variables between regions. Referring
to the Boston Consulting Group product mix matrix, the report
wx
distinguishes between sleeping birds ! , question marks or wild
cats, cash cows, and stars amongst the regions in Europe. This
would explain the difference in findings, assuming East Gothia is
a region with a low science, technology, and innovation profile.
However, this assumption is problematic given the nonpatent
Ž.
citation data compiled by Schild no date and the fact that the
fastest growing science park in Europe is situated in this region.

()
M. MeyerrResearch Policy 29 2000 409–434 425
disclosure in multiple countries, Swedish patent ex-
aminers prefer to cite an English-language patent to
a Swedish or other-language patent. Only if they
have to choose between a Swedish or other non-En-
glish language reference, would they prefer the
Ž
Swedish reference Leif Karlsson, personal commu-
.
nication . This is different in Germany though. Here,
the examiner prefers to cite German-language docu-
Ž.
ments Rainer Bertram, personal communication .
Ž
Provided this is the cause for the different results no
local citations in the Swedish study, local proximity
.
in the US research , one could interpret this as a
further demonstration of the mediated character of
the link patent citations established between science
and technology. 27
6. Discussion
The evidence collected in the study supports the
view that there is a general connection between
science and technology, but it points out that citation
linkages hardly represent a direct link between cited
paper and citing patent. This general conclusion does
not imply that scientific activities would not be of
importance to technological development. The cases
show the opposite. Scientific findings are important
background knowledge playing an important indirect
rather than direct role. The inventors interviewed
pointed out that their inventions were often based on
general experience in research and teaching. General
experience can be seen as one component of tacit
knowledge, which is conveyed chiefly through per-
sonal interaction in a scientific andror technological
environment. In science-related fields, the ability to
absorb tacit knowledge seems to be linked to a
certain degree of scientific education. Most of the
inventors interviewed had some kind of researcher
training. This underlines how important scientific
training is for inventive activity in certain fields.
Some of the inventors in this survey have intermedi-
ary positions between science and technology in the
sense that they are employed and financed by an
27 Furthermore, one should note differences even within Europe.
The patent offices in Italy, France, and Switzerland are nonexam-
ining and, therefore, provide registration only. Examination was
carried out just for certain fields, such as textiles and clock works.
Nowadays, there is no examination in these fields anymore, either.
industrial enterprise and have an academic position
in the university system. This also emphasizes the
significance scientific activities have in supplying
human capital. In this context, the cases support
prior work by Narin et al., suggesting that patent
citations to scientific literature are a general indicator
of science-relation of a field legitimizing basic re-
search activities. Patent citations do reflect the de-
gree to which papers are part of the technological
state-of-the-art. However, it should be borne in mind
that by enumerating publishing and patenting activi-
ties, patent citation studies capture just one aspect of
Ž.
knowledge, namely what Nelson 1998 calls the
‘body of understanding’. They do not reflect knowl-
edge about the ‘body of practice’, which is related to
the design, development, production, marketing, and
use of a particular product model or a specific
Ž.
product line see also Pavitt, 1998b .
As pointed out earlier, Narin et al. appear to have
a basically linear understanding of the innovation
process, postulating ‘science dependence’ of certain
Ž.
technologies Carpenter and Narin, 1983 . The cases
have shown how to interpret this kind of terminology
by pointing to the increasingly intertwined relation-
ship between scientific and technological activities.
As some of the cases presented here have shown,
technology can drive science, too. Some of the cases
have shown that technological developments as indi-
cated by patents take place before their scientific
rationalization as indicated by research papers. This
is better reflected by the two-branched model of Rip
presented earlier, which stresses more the interplay
between the ‘scientific’ exploration branch and the
‘technological’ exploitation branch, allowing also for
a post-rationalization of technological results by sci-
entific scrutiny and technology setting the research
agenda of science. Maybe, it would be more sensible
to use another, less ‘pushy’ term to advertise or
Ž.
describe the nature of a citation link. Schmoch 1993
Ž.
p. 193 used the phrase ‘science–technology inter-
action’ which appears to be a much better descrip-
tion of the actual relationship.
‘Linkage bibliometricians’ look at statistical ag-
gregates, not at single cases. Of course, the 10 cases
presented here are just an entrance point and by no
means sufficient to show any overall effects. How-
ever, the interviews with patent experts illustrate the
mediated character of the connection yet another

()
M. MeyerrResearch Policy 29 2000 409–434426
time. They indicate that nationally different patenting
practices are a major reason for differences in cita-
tion frequencies between countries. The interviews
also provide a possible explanation as to why patent
citation studies indicate local proximity of citations
Ž
in one country and not in another no local citations
in the Swedish study, local proximity in the US
.
research . Again, nationally different examining
practice is an explanation. One could interpret this as
a further demonstration of the mediated character of
the link patent citations establish between science
and technology. It also shows once more that one
should be extremely careful in applying these indica-
tors. The nonexistence of references in patent docu-
ments citing ‘local’ research papers should not be
misinterpreted as an illustration of the nonexistence
of a link between local science and technology. 28
7. Conclusions
The cases are consistent with previous work char-
Ž.
acterizing or at least implying patent citation links
to scientific literature as being of a highly mediated
nature. The cases also show that the citation links do
not indicate science-dependence of technology, but
should be taken as an indication of the multifaceted
interplay between science and technology. This im-
plies that one should not use the linear model to
interpret patent citation data. However, such an inter-
pretation does not invalidate the method of patent
28 Beware also of similar misinterpretations with regard to ‘fore-
ign dependence’ of technologies. The data from the patent expert
interviews indicate that at least in the case of Sweden, patent
citations would suggest that technology patented is more foreign-
dependent than it actually is due to the aforementioned tendency
of patent examiners to prefer corresponding English-language
patents to Swedish ones. This finding contradicts to some extent
Collins and Wyatt who accuse applicants and examiners of na-
Ž11.
tional chauvinism see . However, Collins and Wyatt have not
dealt with patents that originated in Sweden. While the author
Ž
would not be prepared to deny categorically the existence or
.
possibility of Swedish national chauvinism, it has to be said that
there is no evidence of it in Swedish patenting practice. To the
contrary, Sweden’s patenting practice is remarkably cosmopolitan.
As there is not such a tendency in Germany, one might speculate
whether this situation is a result of Sweden’s small country
position.
citation analysis as a policy tool. Policy-makers can
still use this method to illustrate the science-relation
of technological fields and have a mediated justifica-
tion for basic research expenditure in certain fields.
However, there are many and perhaps better justifi-
Ž
cations for public funding of basic research see e.g.,
.
Martin and Salter, 1996; Pavitt, 1998a . Patent cita-
tion analysis as a policy tool might have more useful
applications than an ex-post justification of publicly
funded basic research. Its focus should be changed.
Patent citations of NPRs or science references and
Ž.
their frequencies can at least indirectly indicate
varying intensities of interrelation between science
and technology, different from area to area. In some
Ž
fields, it is stronger more citations of publications
.Ž.
per patent , in other fields, weaker fewer . This
would argue for sector-specific technology transfer
policies. Different fields have a different nature of
interaction. This might result in different forms of
exchange and may require different transfer mecha-
nisms. In combination with other methods, citation
analysis may give some hints as to what require-
ments such a mechanism has to meet in a certain
science–technology field. As it also points to fields
in which there is little interaction between science
and technology, it may induce further investigations
as to whether more exchange between academic and
industrial research could enhance the further devel-
opment of a technology.
What one should not do with this method is to
make comparisons and draw conclusions about the
effectiveness and efficiency of knowledge transfer
amongst fields. Even from the limited number of
cases presented here, it is possible to see indications
for different patterns of citation frequencies accord-
ing to fields. Other studies confirm this on a broader
Ž
empirical basis e.g., Meyer-Krahmer and Schmoch,
.
1997 . In addition, one must bear in mind that not all
‘technology’ is patented. Software, for instance, has
been a field with little patenting. Sectoral trends,
such as ‘know-how’ and ‘speed to market’ as the
preferred method of attaining competitive advantage,
may distort a comparison.
Future research should therefore focus on further
developing a more enlightened interpretation of
patent citations and take the limitations of the data
more into account. In addition, future research should
concentrate on how to utilize patent citations analy-

()
M. MeyerrResearch Policy 29 2000 409–434 427
sis — in spite of its mediated nature — as a tool of
strategic science and technology policy planning.
Acknowledgements
I want to thank the inventors Olle Nilsson, Lars
Thylen, Franz Hofmann, Harald Fuchs, Claudius Ko-
´
rmann, Mayk Kresse, Bror Morein, Gunnar Brandt,
Rudiger Nass, and Stefan Miltenyi for their participa-
¨
tion in this survey. I am grateful to Rainer Bertram,
Kristiina Gronlund, Gary Jordan, Leif Karlsson, and
¨
Hans-Hermann Zitt for their information on patent
practice. First of all, thanks to Sylvan Katz whose
comment on another article initiated the research that
has led to this paper. For comments on earlier drafts,
I want to thank John Clark, Chris Freeman, Gary
Jordan, Tarmo Lemola, Terttu Luukkonen, Ben Mar-
tin, Francis Narin, Keith Pavitt, Olle Persson, and
Ulrich Schmoch as well as the participants of the 3rd
Nordic Workshop in Bibliometrics in Oslo, Norway.
I am especially grateful to Leif Hommen for his
comments and his editorial advice. Two anonymous
referees have also greatly contributed to this paper.
Finally, many thanks to Miquel Aguado, Peter Board,
and Sami Mahroum for comments on a related arti-
cle. Earlier drafts of this paper were written during
my stay at the Research Policy Group, Tema T,
Linkoping University, Sweden, and the VTT Group
¨
for Technology Studies, Espoo, Finland. I am grate-
ful for their hospitality. Research in Sweden was
supported by ‘The Bank of Sweden Tercentary
Foundation’. My stay in Finland was financed by the
Centre of International Mobility, CIMO. This article
will contribute to my doctoral studies which are
kindly supported by Technopolis.
Appendix A. Case summaries
A.1. Case 1: US patent no. 5,566,197 — tunable
gain coupled laser deÕice
The basic idea of this patent is to use quantum dot
or quantum wire structures to prevent carrier diffu-
sion and enhance gain. According to Olle Nilsson,
the inventor, the properties used in the device
patented have been well known to specialists for
many years. The inÕentor stresses the importance of
tacit knowledge for this inÕention 29 when he says
this invention is based on the general knowledge he
had acquired in this field by doing research and by
teaching for many years. This included reading the
scientific literature, visiting conferences, etc. This
description by the inventor underlines the idea that
scientific activities are important in terms of inven-
tive activity. However, the references Nilsson cited
do not reflect the major inputs to the invention at all.
As for the patents cited, the inventor was unaware of
them during the process of invention. As he ex-
plains, citations are more the result of the examina-
tion procedure than they reflect the origin of the
inÕention, as pointed out in the background section.
‘‘They came up in the screening process for the
patent application and imposed limits on what could
be included in the application.’’
Olle Nilsson’s patent has six NPRs. He divides
them into three different groups according to type of
Ž.
publication and purpose of the citation: 1 advanced
level textbooks: their purpose is to support his de-
Ž.
scription of the general field; 2 journal articles,
conference and departmental working papers: Nils-
son used them to support his suggestion to use
certain, well-known quantum phenomena for carrier
Ž
diffusion prevention The inventor knows the authors
of the cited papers, but has never collaborated with
.Ž.
them. ; 3 NPR of restricting character found during
the prosecution: it describes a similar pumping and
tuning system for dye lasers. Here, no diffusion
problem exists. According to the inventor, he was
unaware of this when he got his idea to use a similar
scheme for waveguide semiconductor lasers.
Nilsson chooses a reference to make writing his
patent application easier, i.e., he looks for the most
appropriate references to distinguish his claims from
29 Case studies often rely on interview data. This study is no
exception. This section reports on the history of the inventions as
seen by the inventors. However, sometimes, it seems appropriate
to insert explaining or interpretative remarks. In order to distin-
guish between reported account by the inventor and interpretation
by the author of this paper, italics indicate those parts in this
section that go beyond a mere depiction of the inventor’s views
and experiences.

()
M. MeyerrResearch Policy 29 2000 409–434428
‘prior art’. This indicates that describing the cogni-
tiÕe origins of his inÕention is hardly reflected in the
citations looked at.
A.2. Case 2: US patent no. 5,367,274 — quantum
waÕe guiding switch
Lars Thylen, the inventor, got the idea leading to
´
this patent during his summer vacation while he was
reading an article on waveguides in Physics Today,
a journal for physicists. He realized that there were
similarities to what he was doing in optics. ‘‘I had a
look at it just for fun,’’ says Thylen nowadays. ‘‘I
´
knew the problems of classical switches.’’ The in-
vention patented is basically a transfer of knowledge
from one technological area to another. To see what
has been published, Thylen did some literature re-
´
search. He also compiled a number of patents.
He considers papers more important than patents
in terms of input to the invention. Thylen remembers
´
just one of the patents cited and this only because it
was incoherent. He simply believed what these re-
searchers patented would never work. One reason for
this is the motive that inventors have for using
references to patent documents, namely to support
the inventor’s claims for novelty and uniqueness of
the invention. In other words, they are citing patents
to justify claims. The patents cited, however, ‘‘did
not really influence the work we really did’’.
Apart from 11 patent documents, Thylen cites 12
´
NPRs: nine journal articles, two conference papers,
and one book. The inventor is acquainted with four
Ž
of the cited authors. With two of them Dagli and
.
Datta , there has been a history of cooperation. In the
inventor’s view, one of the journal articles and the
conference paper had given some input to the inven-
tion. Thylen found some early references in an arti-
´
cle authored by a collaborator, who also functioned
as an external examiner. The other paper that con-
tributed some input to the invention was presented
by another collaborator. The other authors Thylen
´
Ž.
cited Petroff and Sakaki were known from confer-
ences. According to the inventor, in general, he cited
papers to give an appropriate background and to
verify his claims. Sometimes the inventor was not
able to recognize the other authors and papers in
detail anymore.
A.3. Cases 1 and 2: InÕentors’ history
Thylen and Nilsson know each other. There has
´
been a long history of cooperation and collaboration.
As a professor, Thylen has been with the Department
´
of Electronics at KTH — the Royal Institute of
Technology in Stockholm — since 1992. Prior to
this, he was with Ericsson, the Swedish telecom
company, working in the area of optics — in partic-
ular, fibre-optic components, digital electronics, and
wave guiding. He decided to leave Ericsson after 20
years for KTH, where he was an adjunct professor.
Nilsson is an adjunct professor with the Electrical
Engineering Department at CTH — Chalmers Tech-
nical University in Gothenburg — as well as at
Thylen’s department. He was Thylen’s predecessor
´´
in the professorship before he went to Ericsson.
This brief personal background does not tell any-
thing about how the ideas for both inventions as such
were formed and materialized, but it sheds some
light on the personal knowledge transfer that has
taken place between two academic research institu-
tions — CTH and KTH — and Ericsson. If one is
more interested in actual knowledge transfer, looking
at assignee–inventor links seems to be a better choice
than tracing back citation linkages.
The two cases are also a good example for how
difficult it is to distinguish between scientific and
technological activities. Both of the inventors inter-
viewed were in so-called adjunct professorship posi-
tions. An adjunct professor here means a researcher
employed by a corporate R&D organization who
spends a considerable amount of time, usually 20%,
sometimes 50%, working and teaching at a univer-
sity department. For this case, it is hard to tell what
was science and what was technology.
A.4. Case 3: US patents no. 5,418,197 and 5,420,083
— SiC whisker and particle reinforced ceramic
cutting tool material and whisker and particle rein-
forced ceramic cutting tool material
Both patents deal with whisker and nanosize par-
ticle reinforced ceramic cutting tools. The difference
between both patents is the use of SiC in one of
them. Both patents contain citations of patents as
well as of the scientific literature. The patents have
five citations of patents and one NPR in common.
The general patent contains three further references

()
M. MeyerrResearch Policy 29 2000 409–434 429
to patents and another citation of a basic research
paper, while the SiC patent has two additional patent
references. However, the different patents and papers
did not really affect the process of invention.
‘‘The patents are the result of my experience in
this field,’’ says Gunnar Brandt, the inventor. He
works in the R&D department of Sandvik, the
Swedish materials company. He has been involved in
research on nanowhisker and particle ceramics since
the beginning of the 1990s. Though he knows some
of the authors and inventors, it is his own prior work
that stimulated the invention, which illustrates the
importance of tacit knowledge in the inÕention pro-
cess. His own patents that are cited belong to a
related field. ‘‘The inventions have simply grown
out of what I was doing then,’’ explains Gunnar
Brandt. He also states that scientific activities of
academic institutions are important: ‘‘We look at
what the universities are doing’’. Despite the signifi-
cance of the scientific activities as reflected in re-
search papers, the papers cited do not reflect any
input into the invention. He says, ‘‘I found them
during my work in this project to compare my
invention to what used to be state-of-the-art then’’.
A.5. Case 4: US patent no. 5,603,958 — pharma-
ceutical carriers
This patent deals with the use of a special drug-
carrying nanoparticle. The patent cites three patent
documents and two articles. In this particular case,
the patent citations could be seen as indicators of
familiarity and collaboration. Two of the cited
patents are from inventors known to Bror Morein,
one of the two inventors. All of the scientific re-
search papers are by authors Morein knows. One of
them is written by his co-inventor, who works in the
same lab. The paper by his co-inventor resulted from
the same project that led to this patent. The article
gives a general overview about the various lipid
components involved, while the other paper de-
scribes in a more detailed manner the chemistry of
certain components important to the invention. The
main purpose for citing articles as well as papers was
to give an overview of the state-of-the-art in this
particular area. Scientific literature does play a role
in this respect. It has the function to support the
noÕelty claim of the patent by describing prior art.
HoweÕer, the importance of NPRs Õaries from field
to field. Says Morein, ‘‘we are not closely following
the drug delivery literature. Therefore, we look more
closely at the vaccine papers’’. Here, we actually
find a certain kind of relatedness between patent and
cited literature. The relationship established by cita-
tion indicates personal and subject-related familiar-
ity.The patent is completely based on intra-mural
research. ‘‘All the work was done in my lab,’’
Morein says. Though Morein, based with his group
at the Agricultural University of Sweden, is exploit-
ing the patent, the assignee is a UK company — the
Ž.
British Technology Group BTG . BTG is a com-
pany originally founded to take care of UK univer-
sity-generated inventions. Now they are operating
worldwide. With 310 US patents assigned to them,
‘‘they are the leader’’. BTG also collaborates with
Morein’s group in other related fields, namely vac-
cines. ‘‘They have a broader customer base for
carrier systems, and so we let them exploit the
patents.’’
This case illustrates an interesting situation a small
country might face. A leading research group in an
advanced field does not seem to have a potent
commercial counterpart within the country and feels
itself forced to collaborate with a foreign firm. As
linkage bibliometric research often neglects small
open economies in its coverage, it would be of
interest to find out whether there is a higher ten-
dency of small country-originated patents to be as-
signed to foreign multinational companies.
A.6. Case 5: US patent no. 5,427,767 — nanocrys-
talline magnetic iron oxyde particles-method for
preparation and use in medical diagnostics and ther-
apy
The applicant of the patent is the ‘Institut fur
¨
Ž.
Diagnostikforschung’ IDF . IDF is a basic research
institute funded by Schering, a leading pharmaceuti-
cal company in the contrast media market, and since
1990, a so-called An-Institut of the Freie Universitat
¨
Berlin. The patent is about a method for preparation
and use of nanocrystalline magnetic iron oxide parti-
cles in medical diagnostics and therapy. The patent
contains 25 references citing to other patent docu-
ments. There are no citations to the scientific re-
search literature. According to Kresse, there are many
scientific research papers as well as patents in this

()
M. MeyerrResearch Policy 29 2000 409–434430
field. In particular, he has more than 600 patents in a
personal database. Asked why the patent does not
contain any citations to the literature, he pointed out
that in this case, the recommended presentation of
‘prior art’ is most efficiently done by referring to the
patents only. ‘‘The patents are already covering the
enormous number of scientific publications. This is
due to the fact that, although the field is science-
based, most of the research is done by companies.
Nevertheless, scientific research papers give valuable
hints in special cases.’’
Kresse divided the cited patents into three classes:
Ž.
1 standard patents that show the novelty of the
Ž.
invention by referring to prior art, 2 patents of
Ž.
collaborators, and 3 patents of competitors. Eight
of the 16 patents belong to the first group of patents,
while the remaining eight actually list inventors
whom Kresse could identify either as collaborators
or as competitors. There are also exchanges of infor-
mation with competing researchers. Those usually
take place at conferences. There, one is informed by
others about what they write and publish or patent.
Collaborations naturally go further. However, in this
particular case, they have not had an effect on the
invention. ‘‘On the contrary,’’ says Kresse, ‘‘this
patent is a real in-house product, where the underly-
ing research work is done by PhD students. Thus,
this patent is a good example of a side-line of
in-house research.’’
This case indicates the tacit knowledge transfer
Ž.
taking place in post graduate education between
industrial sector and academia. It is indeed a good
demonstration for university–industry collaboration.
However, it is not to be detected by NPR citation
measures. Another problem here seems to be the
location of distinct spheres of ‘science’ and ‘technol-
ogy’. IDF is a university institute, but it is financed
by and affiliated to a business firm. It is difficult to
draw the line between scientific research and techno-
logical development.
A.7. Case 6: US patent no. 5,559,353 — integrated
circuit structure haÕing at least one CMOS-NAND
gate and method for the manufacture thereof
This patent describes a novel three-dimensional
arrangement of MOS transistors leading to higher
density. The patent also includes a manufacturing
method of the circuit structure by the epitaxal defini-
tion of the layer sequences. It resulted from in-house
research of Siemens, the German electronic engineer-
ing company. According to Franz Hofmann, one of
the four inventors, the invention did not involve
researchers outside the nanoelectronics group at
Siemens Research Center in Munich. The patent
contains six references to other patent documents
and three NPRs. In this particular case, the various
citations document the collaboration of Siemens’
researchers and patent specialists.
‘‘When we had the idea for this patent, we went
to our patent specialists,’’ Hofmann says. Together
they wrote the patent application. The patent depart-
ment did some research to find both patents as well
as publications related to the invention. As the inven-
tor explains, the aim was to find examples for the
prior state of technology and thereby document the
novelty of the invention. According to Franz Hof-
mann, all of the references to other patents and two
of the three citations to scientific research papers
were found by the patent specialists and are not
related to the invention process. The last reference
cites an author who was personally known to the
group. However, the invention has been made on the
basis of general knowledge and experience and is not
so much due to particular inputs of external scien-
tists.
This case illustrates the legal function of patent
citation as discussed in earlier sections of the paper.
References are cited to meet the special criterion of
novelty. An interesting finding is furthermore the
strong role the patent department seems to play. This
underlines the social nature of the patenting process
and shows how important the interaction between
inventors and patent specialists are. The case also
demonstrates how important local collaboration in
R&D is — ‘local’ here understood as ‘in-house’
rather than ‘extra-mural, but nearby’.
A.8. Case 7: US patent no. 5,298,760 — perfor-
mance of location-selectiÕe catalytic reactions with
or on the surfaces of solids in the nanometer or
subnanometer range
This patent is about a chemical surface modifica-
tion in the nanometer and subnanometer range by
means of scanning-probe microscopy. The process is
used for information storage. Harald Fuchs, nowa-
days professor at the ‘Physikalisches Institut’ in

()
M. MeyerrResearch Policy 29 2000 409–434 431
Munster, Germany, is one of the two inventors. He
¨
was then working at BASF. His co-inventor was his
assistant at that time. The patent resulted from a
basic research project at BASF. It was part of the
company’s contribution to the ‘Ultrathin Films Pro-
gram’ of the BMBF, the German Federal Ministry
for Education and Sciences, Research and Technol-
ogy.
The patent contains one reference to a patent
document. There are four further references listed,
three of which are scientific research papers. The
remaining reference is to a technical disclosure bul-
letin of another company. Only one of the research
papers documents a direct contact to another re-
searcher at BASF. Fuchs holds some other patents
and has co-authored a number of articles together
with his co-inventing colleague. Fuchs agrees that
the citations to scientific research papers document
the basic research character of his project at the time
it was undertaken. However, the citations themselves
were the result of literature and database research
done afterwards and independently of the original
invention. In Fuchs’ view, ‘‘the vision for this inven-
tion was the result of my own work. I just had to
substantialize and calibrate it against the state-of-the-
art.’’
This case shows a direct link insofar as science
and technology ‘happened’ in the same project, which
is illustrated in a science citation. Here, scientific
papers and patented technology resulted from the
same effort. Again, this case illustrates the close
embrace of science and technology. The same re-
searchers delivered papers as well as technological
results. But despite the very close and direct link,
citation links do not seem to reflect the cognitive
origin of the invention.
A.9. Case 8: US patent no. 5,543,289 — methods
and materials for improÕed high gradient magnetic
separation of biological materials
This patent discloses improvements in procedures
and materials for high gradient magnetic separation
Ž.
HGMS of biological materials. It also contains a
method to conduct HGMS. The patent cites 25 US
and two other patents. Five papers are cited. As
Stefan Miltenyi, the inventor, explains, the usually
large number of US patents cited is due to the US
patent law that makes it possible to proceed against
patents and their claims at any time. ‘‘Everything
that could be somehow relevant needs to be in-
cluded.’’
This case also illustrates the influence different
areas of technology have on patenting. Miltenyi
commissions Californian lawyers to take care of his
patenting activities. ‘‘They are specialized into bio-
technology.’’ The lawyers themselves add references
to make the application ‘water-proof’. The refer-
ences cited do not indicate direct input into the
invention. They are primarily there to relate the
invention to prior art. In this particular field, as much
as possible is patented. ‘‘People patent before they
publish, if they do so at all.’’
This case illustrates the influence special fields
can have on patenting. This case is undoubtedly
taken from a highly science-based field, and yet our
inventor’s observation is that with respect to publica-
tion and patenting, it shows a reverse chronological
pattern. One of the reasons for this particular practice
can be seen in the legal implications, as Cambrosio
Ž.
et al. 1990 pointed out. While in other fields, small
and medium-sized companies or individual inventors
file not infrequently themselves, biotech SMEs seem
to rely on specialist patent attorneys. The much
higher patent citations may also reflect the different
examination practices between Europe and the US.
In this case, a German inventor commissioned US
patent attorneys instead of contacting German attor-
neys first and then filing a disclosure in multiple
countries, including the US. The much tougher legal
Ž.
situation in both the technological field biotech and
Ž.
the country US might have had an amplifying
effect in terms of patent references.
A.10. Case 9: US patent no. 5,470,910 and 5,590,387
— composite materials containing nanoscalar parti-
cles, process for producing and their use for optical
components and method for producing metal and
ceramic sintered bodies and coatings
Both patents deal with nanostructured materials
and their production. While the first patent
Ž.
5,470,910 discloses special composite materials
containing nanoparticles, their fabrication as well as
their use for optical components, the latter one is a
process patent describing a way of producing metal
and ceramic sintered bodies and coatings using
nanocrystalline metal or ceramic powder. The first

()
M. MeyerrResearch Policy 29 2000 409–434432
Ž
patent exclusively lists references to patents five
.
references to US patents, three to other patents . The
second one also contains 13 patents. It also lists three
‘other references’. But even they are related to
patents. Patent abstracts, the related Japanese patent
itself and a search report of the European Patent
Office are included. There are no references to scien-
tific research papers in either one of the two patents.
‘‘This depends on the orientation of the respective
research establishment that carries out research on a
certain topic,’’ says inventor Rudiger Nass of INM,
¨
the Institute of New Materials in Saarbrucken, Ger-
¨
many. According to Nass, applied research — as it
is done at INM — is naturally more patent-oriented
than university-based research. Who exploits the in-
Õentions patented does not seem to haÕe any impact
in this case. Nass did not point out any differences
between the two patents, one of which is assigned to
an affiliate of a multinational chemical company,
while the other patent is exploited by INM itself,
with regard to patenting and citation.
According to Nass, there are basically two cate-
Ž.
gories of citations in patents: 1 citations of patents
and papers made by the inventors themselves, and
Ž.
2 citations due to the examination process. Fur-
thermore, Nass describes the strategic nature of
patents and patent citations by distinguishing two
Ž.
different types of patents: 1 one works in a certain
field and has designed something new, an invention
in its original sense — something which has never
been made before. In this case, the inventor would
not be familiar with the relevant patent literature. He
or she would search for related patents after the
invention was made. However, patents of this kind
Ž.
occur relatively seldom. 2 More likely is a situation
in which the inventor is facing a concrete problem.
To solve it, he or she has to know how to evade and
avoid existing patents. This, however, requires a
rather intimate knowledge of related patents. One has
to consider what there is of available literature.
Scientific research papers become relevant only when
one has to deal with special cases.
A.11. Case 10: US patents no. 5,500,141 and
5,250,207 — magnetic ink concentrate
BASF, one of the large German chemical compa-
nies, holds a number of patents in the field of
nanostructured materials. Here, I look at a patent
Ž.
and its continuation for a magnetic ink concentrate
and a process for its preparation. Claudius Kormann,
one of the inventors, was given the task to ‘‘make
something new out of a 50-year-old area’’. The
patents are the result of in-house cooperation. Apart
from Kormann’s superior, the other inventors were
members of a different department. They gave valu-
able input from the application side of superparam-
agnetic solid particles.
The original patent contains 14 references to other
patents. Furthermore, it refers to three NPRs, namely
patent and scientific abstracts. The main purpose of
the cited patents and abstracts is — not surprisingly
— to document the state-of-the-art. Most of the
documents referred to were not known to the inven-
tor previously. One of them, however, does illustrate
in-house cooperation. Kormann knew about this par-
ticular patent since it belonged to the colleagues he
collaborated with.
Looking at both patents, the original one and its
continuation, show other reasons for which patents
and papers are cited. The continuation of the original
patent contains two further citations to US patents.
They were included because they restrict the original
claims made in the older patent. Referring to the
‘A–X–Y’ classification of the European Patent Of-
fice, Kormann says, basically ‘‘there are two kinds
of patents: those which summarize prior develop-
ments in fields related to the invention, and those
which make inventors modify their claims’’. Patents
also contain references that were found by the patent
examiner. They are included to specify the claims
and avoid misunderstandings.
According to Kormann, citations made in patents
to scientific literature do not reflect any input of
basic research related to the invention. The case is
quite the opposite. ‘‘Patents in very advanced and
modern fields usually do not contain references to
scientific research papers.’’ Despite a time lag in
patenting of about 18 months, it is science that is
lagging behind.
This case shows again the peculiarity of technical
fields by pointing out that technology can guide
science in certain areas: first, invention, then scien-
tific rationalization. Another finding one can draw
from the case is that differences in sciento-techno-
Ž
logical practice occur within the firm see case 7 that
.
follows a different pattern .

()
M. MeyerrResearch Policy 29 2000 409–434 433
From what was reported by one of the inventors,
one can see some potential a co-inventor analysis
would hold. The list of co-inventors illustrates the
Ž
collaborative efforts between different aspects and
.
representatives of technology that were necessary to
achieve the inventive result. Another issue then is the
question to what extent one could trace back the
co-inventor affiliation down to an intra-organiza-
tional level, such as departments.
References
Anttinen, P., Lounasmaa, J., 1997. Nanotechnology in the USA.
Report for the Technological Development Centre of Finland
TEKES, Helsinki.
Arora, A., 1991. The transfer of technological know-how to
developing countries: technology licensing, tacit knowledge,
and the acquisition of technological capability. PhD disserta-
Ž.
tion, Stanford University, see David 1993 .
Bachmann, G., 1995. Nanotechnology: technology analysis. VDI-
Technologiezentrum Physikalische Technologien, Dusseldorf,
¨
Germany.
Bare, R., 1981. Results of a statistical study of the references cited
´Ž.
in the search reports established by the EPO January 1981 .
Ž.
World Patent Information 3 2 , 56–60.
Bell, M., Pavitt, K., 1993. Technological accumulation and indus-
trial growth: contrasts between developed and developing
Ž.
countries. Industrial and Corporate Change 2 2 , 157–210.
Braun, T., Schubert, A., Zsindely, S., 1997. Nanoscience and
nanotechnology on the balance. Scientometrics 38, 321–325.
Budworth, D.W., 1996. Overview of activities on nanotechnology
and related technologies. Institute for Prospective Technology
Studies of the European Commission, Joint Research Center,
IPTS Technical Report Series, IPTS-JRC, Seville.
Cambrosio, A., Keating, P., MacKenzie, M., 1990. Scientific
practice in the courtroom: the construction of sociotechnical
identities in a biotechnology patent dispute. Social Problems
Ž.
37 3 , 275–293.
Campbell, B., 1989. Generalists, practitioners, and intellectuals:
the credibility of experts in English patent law. In: Smith, R.,
Ž.
Wynne, B. Eds. , Expert Evidence: Interpreting Science in the
Law. Routledge, London, pp. 210–236.
Campbell, R.S., Nieves, A.L., 1979. Technology Indicators Based
on Patent Data. The Case of Catalytic Converters — Phase I
Report, Design and Demonstration. Batelle Pacific Northwest
Laboratories, n.p.
Carpenter, M.P., Narin, F., 1983. Validation study: patent cita-
tions as indicators of science and foreign dependence. World
Ž.
Patent Information 5 3 , 180–185.
Collins, P., Wyatt, S., 1988. Citations in patents to the basic
research literature. Research Policy 17, 65–74.
David, P.A., 1993. Intellectual property institutions and the panda’s
thumb: patents, copyrights, and trade secrets in economic
Ž.
theory and history. In: Wallerstein, M.B. et al. Eds. , The
Global Dimensions of Intellectual Property Rights in Science
and Technology: a Conference, January 7–8, 1992. National
research Council, National Academy Press, Washington, DC,
pp. 19–61.
Dosi, G., 1988. Sources, procedures and microeconomic effects of
innovation. Journal of Economic Literature 26, 1120–1171.
European Commission, 1997. The second European report on
science and technology indicators. Report EUR 17639 EN,
Chap. 7, in particular, p. 359ff.
Ganguli, P., Blackman, M.J.R., 1995. Patent documents: a multi-
Ž.
edge tool. World Patent Information 17 4 , 245–256.
Grupp, H., Schmoch, U., 1992. Perceptions of scientification of
innovation as measured by referencing between patents and
papers: dynamics in science-based fields of technology. In
Ž.
Grupp 1992 pp. 73–128.
Hicks, D., 1995. Published papers, tacit competencies and corpo-
rate management of the publicrprivate character of knowl-
Ž.
edge. Industrial and Corporate Change 4 2 , 401–424.
Hilgartner, S., 1990. The dominant view of popularization: con-
ceptual problems, political uses. Social Studies of Science 20,
519–539.
Howells, J., 1996. Tacit knowledge, innovation and technology
transfer. Technology Analysis and Strategic Management 8
Ž.
2 , 91–106.
Iversen, E.J., 1998. Knowledge bases and interactions in the
Norwegian system: a patent share and citation analysis. Dis-
cussion draft, dated 8 May 1998, for ‘STEP rapportrreport
1997’. Presented at the 3rd Nordic Workshop in Bibliometrics,
Oslo, 3–4 August 1998.
Jaffe, A.D., Traitenberg, M., Henderson, R., 1993. Geographic
localization of knowledge as evidenced by patent citations.
The Quarterly Journal of Economics, 577–598.
Kuusi, O., 1994. Materiaalit murroksessa. VATT, Helsinki.
Malsch, I., 1997. Nanotechnology in Europe: Experts’ Perceptions
and Scientific Relations between Sub-areas. Institute for
Prospective Technological Studies, Seville.
Martin, B., Salter, A., 1996. The relationship between publicly
funded basic research and economic performance. Science
Policy Research Unit, Brighton.
Marquis, D.G., Allen, T.J., 1966. Communication patterns in
applied technology. American Psychologist 21, 1051–1060.
Meyer, M., Persson, O., 1998. Nanotechnology — interdisci-
plinarity, patterns of collaboration and differences in applica-
Ž.
tion. Scientometrics 42 2 , 195–205.
Meyer-Krahmer, F., Schmoch, U., 1997. Chemistry, information
technology, biotechnology, and production technology: a com-
parison of linking mechanisms in four fields. Fraunhofer Insti-
tute for Systems and Innovation Research, Karlsruhe, Ger-
many.
Ž.
Narin, F., 1994. Patent bibliometrics. Scientometrics 30 1 , 147–
155.
Narin, F., Noma, E., 1985. Is technology becoming science?.
Ž.
Scientometrics 7 3–6 , 369–381.
Narin, F., Hamilton, K.S., Olivastro, D., 1995. Linkage between
agency supported research and patented industrial technology.
Ž.
Research Evaluation 5 3 , 183–187.
Narin, F., Hamilton, K.S., Olivastro, D., 1997. The increasing

()
M. MeyerrResearch Policy 29 2000 409–434434
linkage between U.S. technology and public science. Research
Ž.
Policy 26 3 , 317–330.
Nelson, R., 1998. Different perspectives on technological evolu-
Ž.
tion. In: Ziman, J. Ed. , Technological innovation as an
evolutionary process. Cambridge Univ. Press.
Noyons, E.C.M., van Raan, A.F.J., Grupp, H., Schmoch, U.,
1994. Exploring the science and technology interface: inven-
tor–author relations in laser medicine research. Research Pol-
icy 23, 443–457.
Pavitt, K., 1984. Sectoral patterns of technical change: towards a
taxonomy and a theory. Research Policy 13, 343–373.
Pavitt, K., 1987. The objectives of technology policy. Science and
Ž.
Public Policy 14, 182–188, see David 1993 .
Pavitt, K., 1998. Do patents reflect the useful research output of
universities? SPRU Electronic Working Papers series. Paper
No 6. Science Policy Research Unit, Brighton, also published
in Research Evaluation 23, 105–111.
Pavitt, K., 1998b. Technologies, products and organisation in the
innovating firm: what Adam Smith tells us and Joseph Schum-
peter doesn’t. Industrial Policy and Corporate Change 7, 433–
452.
Rabeharisoa, V., 1992. A special mediation between science and
technology: when inventors publish scientific articles in fuel
Ž.
cells research. In Grupp 1992 , pp. 45–72.
Rip, A., 1986. Mobilising resources through texts. In: Callon, M.,
Ž.
Law, J., Rip, A. Eds. , Mapping the Dynamics of Science and
Technology. Macmillan, London, pp. 84–99.
Rip, A., 1992. Science and technology as dancing partners. In:
Ž.
Kroes, P., Bakker, M. Eds. , Technological Development and
Science in the Industrial Age. Kluwer, n.p., pp. 231–270.
Rosenberg, N., 1990. Why do companies do basic research with
their own money?. Research Policy 19, 165–174, see David
Ž.
1993 .
Schild, I., no date, Using patent indicators to map inventive
activity in East Gothia: a methodological exploration with
preliminary results. Work in progress. Linkoping, Tema T
¨
Department of Technology and Social Change, Linkoping
¨
University.
Schmoch, U., 1993. Tracing the knowledge transfer from science
to technology as reflected in patent indicators. Scientometrics
Ž.
26 1 , 193–211.
Schmoch, U., Kirsch, N., Lay, W., Plescher, E., Jung, K.-O.,
1991. Analysis of technical spin-off effects of space-related
R&D by means of patent indicators. Acta Astronautica 24,
353–362.
Tolles, W.M., 1996. Nanoscience and nanotechnology in Europe.
Nanotechnology 7, 59–105.
van den Belt, H., 1989. Action at a distance: A.W. Hofmann and
Ž.
the French patent disputes 1860–1863 , or how a scientist
may influence legal decisions without appearing in court. In:
Ž.
Smith, R., Wynne, B. Eds. , Expert Evidence: Interpreting
Science in the Law. Routledge, London.
van Vianen, B.G., Moed, H.F., van Raan, A.F.J., 1990. An
exploration of the science base of recent technology. Research
Policy 19, 61–81.
Walker, R.D., 1995. Patents as scientific and technical literature.
The Scarecrow Press, Metuchen, NJ.
Whitley, R., 1985. Knowledge producers and knowledge acquir-
ers: popularisation as a relation between scientific fields and
Ž.
their publics. In Shinn, T., Whitley, R. Eds. , Expository
Science: Forms and Functions of Popularisation, Sociology of
the Sciences Yearbook, Vol. 4. Reidel, Dordrecht and Boston,
MA, pp. 3–28.
Yin, R.K., 1994. Case study research: design and methods, 2nd
edn. Sage, Thousand Oaks.