Scientometrics, VoL 28. No. 1 (1993) 89-110 MEASURING NATIONAL OUTPUT IN PHYSICS: DELIMITATION PROBLEMS E. J. RINIA,* C. DE LANGE,* H. F. MOED** * Foundation for Fundamental Research on Matter (FOM), P.O. Box 3021, 3502 GA Utrecht (The Netherlands) ** Centre for of Science and Technology Studies (CWTS), University of Leiden, Wassenaarseweg 5~ P.O. Box 9555, 2300 RB Leiden (The Netherlands) (Received March 5, 1993) In a study of the Dutch publication output in physics we tested methods of delimitating fields by journal categories in the Science Citation Index (SCI) compared to the classification of individual publications into subfields in the subject specific database Physics Briefs (PHYS). Different methods of measuring national scientific output were compared as well. In this paper we report the main findings on these issues, based on a study of six selected subfields in physics 9 The main conclusion with respect to the use of different classification methods is that in most of the selected fields in physics the method which delimitates fields by journal categories yields an incomplete picture of the output of a country. Particularly because this method neglects a considerable number of articles published in general journals. With respect to different methods of counting publications it was corroborated by the Dutch data in Physics Bliefs that: 1. so-called 'integer counted' world shares are very much influenced by the degree of 'internationalisation' and 2. 'first author counting' gives a satisfactory approximation of 'fractional counting'. Citation indicators based on 'first author counting', however, may be distorted in fields with a large fraction of international co-authored publications. Introduction In many science fields disciplinary oriented literature databases exist such as Chemical Abstracts in the field chemistry, Medline, or EMBASE in the biomedical sciences and Physics Briefs (PHYS) or Physics Abstracts (INSPEC) in physics. In most of these databases (sub)fields and topics can be identified on the basis of subject classification systems applied on a paper-by-paper basis: articles can be assigned to fields because subject classification codes corresponding to subfields and research topics are attributed by experts to each single publication. In bibliometrics the Science Citation Index (SCI), which offers both information on publications and citations in all science fields, is widely used. However, in the SCI publications are not classified according to a structured subject-classification scheme. To solve the problem of delimitating fields, an alternative method is used by Scientometrics 28 (1993) Elsevier, Amsterdam - Oxford- New York- Tokyo Akad~miai Kiad6