The results obtained in this study offer us a macroscopic picture of the integration process
since the sample from the Science Citation Index includes all disciplines in natural and life
sciences. The conclusion that the European union is integrating can by no means be
generalised for all scientific disciplines in which research communities are predominantly
organised. It may well be the case that the application of integration measures at the level of
scientific disciplines would show disintegration for some disciplines.16 To look for more
detailed explanations of patterns of collaboration, future research could extend the analysis
presented here by decomposing the collaboration matrix into the lower level of scientific
disciplines. Methodologies based on journal-journal citation reports are readily available do
delineate scientific disciplines using clustering techniques (Leydesdorff and Cozzens, 1993;
Van den Besselaar and Leydesdorff, 1996). Alternatively, one can use exiting classifications
that are available from ISI. 17 Having delineated scientific disciplines, one can apply the
proposed indicators of integration in the same manner as we applied these to the science
systems as a whole. One can then attempt to explain the level of integration of scientific
disciplines as a dependent variable from independent variable that characterise disciplines
(tacitness, specificity, appropriability, fixed costs). From earlier patent studies it has become
clear that differences in spatial concentration of innovative activities in Europe are highly
sector-specific (Breschi, 2000). Similarly, one can expects important differences to exist in
the degree of European integration of different scientific disciplines.
The research agenda outlined above is also expected to contribute to (European) policy design
and policy evaluation. Understanding the determinants of research collaboration from
characteristics of scientific disciplines can be helpful in designing policies that promote
collaboration and mobility at the level of particular disciplines. For example, the current
emphasis of EU science policy on applied knowledge production in Framework Programmes
and the lack of EU funding of basic science, has been criticised at various occasions
(Banchoff, 2002; Pavitt, 2000). This criticism calls for empirical studies that test whether
basic science benefits from supranational networks, and whether applied science more often
emerges from local networks. If so, there may be reasons to re-adjust the orientation of
collaboration and mobility programs of the European Union towards basic science.
16 It is even theoretically possible that all disciplines are disintegrating in islands of collaborating clubs
of countries, while the macroscopic system as a whole is integrating. If specific pairs of countries
would specialise in specific scientific disciplines, but different pairs of countries would specialise in
different scientific disciplines, the integration values of disciplines would show disintegration, while
the aggregated science system could still show a macro process of integration.
17 http://www.isinet.com
17