Scientific research in the natural sciences in South Africa: A scientometric study

As a leading producer of scientific publications on the African continent, South Africa has made remarkable progress. However, attempts are yet to be made to comprehend the empirical reality of scientific production in South Africa. One way to do this is to analyse specific science disciplines (such as the natural sciences), publication outputs and their features. A bibliometric study was undertaken of the publication trends and patterns of South African researchers in the natural sciences from 1975 to 2005 (choosing selected sample years), using the Thomson Reuters’ Web of Knowledge database of selected indexed natural science journals. Characteristics of natural science publications, such as the trends over the years, were revealed as well as the collaborative dimensions involved in the production of scientific papers in these disciplines in South Africa. The connection between collaboration and publication, as well as between collaboration and sectors of authors was evident. The key findings of this study were that authors were based mostly in universities and were collaborative in their research endeavours. In addition, the participation of international collaborators has increased.


Introduction
In the production of scientific publications that are indexed in the Thomson Reuters (ISI) Web of Knowledge, South Africa is a leader on the African continent.During 2000-2004, South Africa's contribution to Africa's indexed scientific publications was 30%, the highest percentage to be achieved by any country on the continent. 1In some of the natural science disciplines South Africa has fared well.During 2000-2004, South Africa topped other African countries in ecology (1187 publications), geosciences (1302 publications) and mathematics (275 publications), but lost its first position to Egypt in other natural science subjects such as chemistry and physics. 1

Analysis and results
The specific characteristics of the publications which appeared in the ISI database for the seven selected years analysed were: the number of publication records, the number of authors involved in the production of these publications, partnering countries, the affiliation sector of authors and their partners, collaboration categories in the production of publications, and the regional origin of international partners.Wherever necessary, the basic characteristics of the publications in the natural sciences were compared with those of publications by South Africans in 'all subjects'.

Publications in the natural sciences
As the sample study shows, South Africa produced 5966 ISI-indexed publications in the natural sciences in the last three decades.When this figure was compared with the total number of publications in 'all subjects', it was 32% of the total South African output (1975-2005).The growth from 415 records in 1975 to 1416 in 2005 was about threefold (341%) in the natural sciences.The year-on-year change, that is, the difference between the current year and the immediate previous year, ranged between 21% and 74%, with the highest percentage of 74% occurring in 2005.Specifically, these changes were 21% in 1980, 45% in 1985%, 40% in 1990, 28% in 1995 and 32% in 2000.A drop in the publication count was seen in 1995, but it recovered in the following years.On average, ISI-indexed natural science publications by South Africans increased at a rate of 40% for every 5-year period of the analysis (Table 1).
Of the 18 466 publications in 'all subjects', 1212 were published in 1975 and 4161 in 2005.The increment between 1975 and 2005 was, as in the case of the natural sciences, threefold.'All subjects' expanded to 343% of the number for the base year 1975.The year-on-year change for 'all subjects' ranged between 4% (in 1995) and 66% (in 2005).The average of these was 41%, which was just 1% over that of the natural sciences.A fall in publications was reported in 1995, for both the natural sciences and 'all subjects'.This fall was faster for 'all subjects' (4%) than for the natural sciences (28%).The F-test (F=0.026)showed that the variation  between the number of publications in the natural sciences and in 'all subjects' was unequal, supporting the finding that the growth trail for the natural sciences was not different from that of the 'all subjects' (Figures 1 and 2).
The average number of authors who worked together in producing publications rose from 2.24 per publication in 1975 to 5.11 in 2005, with an aggregate average of 3.39 for the entire period of analysis.In 1995, the number rose to 138% of the 1975 value, followed by 154% in 2000 and reached an all-time high of 228%.Further analysis revealed that 21% of the papers were the efforts of single authors, 31% of two authors, 22% of three authors, 12% of four authors, 6% of five authors, and the remainder were the work of six or more authors.The proportion of sole-authored papers in 2005 was smaller than that for the entire period of analysis (11% against 21%).A quarter of the publications in 2005 were written by two authors, another 24% by three authors, 17% by four authors, and the remaining 23% by five or more authors.The median value of authors for 2005 was three (Table 2).The number of countries associated with South African scientists can be analysed from two standpoints.The first is the number of countries per publication for the whole sample, and the second is the number of countries per publication for those works resulting from international collaboration.The second standpoint provides a more precise picture than the first.When all papers were taken into consideration, the average number of countries that participated in the production of publications in the natural sciences was 1.49.In 1975, the average number of countries which participated in the production of papers in the natural sciences in South Africa was 1.12, which rose to 1.89 in 2005.Considering 1975 as the base year, the increase in the number of countries was 169%.
Here again, 1995 marked the beginning of a sizable increase of 128%.
Compared to these figures, there were 1.4 countries per publication in 'all subjects' during the period, which was smaller than that of the natural sciences by 0.09 points.However, this difference did not recur between 1975 and 2005 for 'all subjects'.The change in growth of 166% for 'all subjects' was close to the 169% for the natural sciences.
In 1975, there were 2.13 countries that participated in the production of international collaborative papers.By 2005, this number had risen to 2.78, amounting to 131% of the 1975 figure.The average change over the seven selected years was 6%.About 75% of the internationally collaborative papers had brought in scholars from two countries, while 16% brought three countries together, and the remaining 9% brought together four or more countries.
'All subjects' had a participation of 2.56 countries per every internationally produced publication, more than the natural sciences.Between 1975 and 2005, the increase was 130%, about the same as the 134% increase for the natural sciences.The year-on-year average change was also the same for 'all subjects' -5%.The count of countries per publication remained the same for 'all subjects': 74% with two countries, 16% with three, and 10% with four or more countries.
The ISI-indexed publications in the natural sciences by South African researchers and their partners attracted an average of 11.15 citations per item.The highest average -22.31citations per publication -was reported in 1975.By 2000, the count had declined significantly to 8.81, which was only 40% of the citations received in 1975.The citation count in the natural sciences deteriorated from year to year, with an average change between the chosen years of 60%, a loss of 40%.For 'all subjects', the average citation per publication for the period of analysis was 10.53, 0.62 points less than that for the natural sciences.For 'all subjects', the number of citations was 17.26 in 1975 but declined to 9.74 in 2000.The year-on-year change, on average, was 69%, about 9% more than that for the natural sciences.

Sectors of South African authors in the natural sciences
Authors of publications in the natural sciences belong to five major sectors: universities, research institutes, government departments, industries and hospitals.In view of the manageability of records, this analysis focused on the affiliation sectors of up to five authors, the first five South African authors, and the first five international partners.
Most of the authors came from universities, followed by research institutes, government departments, industry and hospitals.As seen in Table 1, the university sector emerged with the highest mean value of 1.65 for the three-decade period.The research institute sector was second with a mean value of 0. Multi-country international collaboration (within international collaboration) ‡ 4 8.9   Personnel in government departments and industries are not expected to produce scientific publications the way academics in universities and researchers in research institutes are.However, the availability of and access to data encourages some to publish material.These authors are self-motivated people who are interested in publicising the findings of key issues for the public.

Authorship and collaboration
About one-fifth of the publications between 1975 and 2005 were produced by single authors while the majority were authored jointly.In the past 30 years, the share of single-author publications has contracted from 38% to 11%.In 2005, the proportion of single-authored papers was reduced to one-third of what it was in 1975.Therefore the share of coauthored papers grew from 66% in 1975 to 89% in 2005.Co-authored publications appreciated to 135% of their original share in 1975, with a year-on-year average growth of 6% between the selected five years.Only in 1980, joint publications dropped to 2.5% of the previous year.A total of 54% of all authors were South African or based in South Africa at the time of publishing their research.This percentage declined to 50% by 2005.
In the natural sciences, collaboration has been increasing.The percentage of collaborative papers for the aggregate period was 79%.In 1975, 66% of papers in the natural sciences involved some form of collaboration.By 2005, this figure had expanded to 89% of all publications.Between the five years, the most significant change was seen in 1990 with 79% of all publications involving collaboration.This change was 120% of the percentage of collaborative papers published in 1975.Again in 2005, there was considerable growth in the number of collaborative papers (89%), i.e. 135% of the 1975 collaborative publications.In about 54% of the publications in the natural sciences (for the whole period), all authors were affiliated with South Africa.Year-wise data showed that in 55% of publications in 1975, all authors were from South Africa; this figure declined to 50% by 2005.
Collaboration can be analysed in a variety of ways.When authors from within a country partner, it is regarded as domestic collaboration, either internal-institutional (authors belong to the same department or institution) or external-institutional (authors belong to different departments or institutions within South Africa).International collaboration means that at least one of the collaborators is from another country.If more than one other country is involved, then it becomes a multi-country international collaboration.Collaborative publications can be domestic, international or both.
In the aggregate period, 67% of the publications were the result of research alliances that occurred within South Africa (domestic collaboration).Overall, 39% of co-authored publications were produced out of international collaborative research.In 1975, international research alliances formed only 17% of the total co-authored publications.This percentage grew steadily over the years, reaching 56% of all co-authored publications in 2005.This figure was more than 2% of the domestic collaboration for the same year.Between 1990 and 1995, the percentage of internationally co-authored papers increased rapidly from 24% to 37%.
The surge in publications during this period is attributed to the changes that occurred in South African science since the late 1980s.For political reasons relating to the apartheid regime, circumstances for scientific growth, publications and co-publications were not conducive until the late 1980s. 10This 'closed off' period was when the South African scientific community was isolated from the international scientific fraternity as a result of a scientific boycott. 5It is only since the early 1990s that, because of democratisation in the post-1994 period resulting in political stability, the conditions for the movement of scientists from abroad began to ease and a collaborative approach towards South Africa by the international scientific community became evident.South African scientists also gained access to international resources including publication opportunities and association with their international peers. 14ith the end of the period of isolation, international journals began to accept the contributions of South African scientists which resulted in an increase in the publications of South African scientists. 5,11other potential factor that influenced the growth of publications in general was the influential funding policies of the government. 15Notably, the number of potential researchers, which in 2006 was 16 000, has also been increasing since 1990. 16thin international collaboration another pattern -multi-country collaboration -has evolved.Multi-country international collaboration made up a good share of the internationally co-authored papers between 1975 and 2005.There has been a significant increase in multi-country international collaboration from 4 papers in 1975 to 228 in 2005.About 13% of internationally authored papers for the whole period involved the participation of more than one other country.Of the multi-country collaborative papers, 64% included the involvement of two other countries, 17% of three other countries, 6% of four other countries, and the remaining 15% of five or more other countries, with an average number of 3.1 countries per every multi-country publication.One paper was authored by researchers from 15 foreign countries -the highest number in collaborative research in the natural sciences during the period of analysis.In this analysis, the degree of collaboration in the natural sciences was measured using the indicators of collaboration -domestic and its types, international, and multi-country collaboration.As per the definition, the presence of all types of collaboration will give a maximum index of four (internal-institutional, external-institutional, international and multi-country international).The mean degrees of collaboration for all papers and collaborative papers for the period of analysis were 0.92 and 1.17, respectively.It is clear from these figures that the presence of collaboration is widespread.In the case of co-authored papers, the index was higher at 1.17.Between the years of analysis the index advanced progressively from 1.04 to 1.32.As for the number of papers with any form of collaboration, the highest collaboration index was in 2005.
Partners of South African authors and their regional origins

Individual countries of overseas partners
The partners of South African researchers in the natural sciences originated from 54 countries, representing all continents (Table 4).Half of the total collaborations were carried out with three countries -USA, England and Germany -and 80% was conducted with 11 countries -the USA, England, Germany, France, Australia, Canada, Italy, the Netherlands, Belgium, Japan and Spain -mostly European and North American countries.The largest contribution -25% -belonged to the USA, followed by England and Germany (12% each).The African countries with which South African scholars worked constituted only 1% of the total collaboration.These countries (in descending order of the share of collaboration) were: Zimbabwe, Ethiopia, Botswana, Namibia, Egypt, Malawi, Swaziland, Tanzania and Zambia.Inter-African collaboration in the natural sciences was mostly evident in environmental sciences/ecology (43% of the total African collaboration) and chemistry (13%).Another 26% was classified by ISI as 'multidisciplinary'.
Collaboration with Asian countries (Japan, China, India, Taiwan, South Korea, Hong Kong and Indonesia) constituted 6% of the whole.In this Asian-South African alliance, Japan, China and India were more prominent with 5% of the contribution.The BRIC countries -Brazil, Russia, India and China -with whom South Africa has established economic collaboration, contributed 19% 39%, 28% and 14% of the total of the four countries, respectively.Collaborations with these countries were mainly in the areas of physics, chemistry, astronomy, geochemistry and geology.

Relation between collaboration and production of publications
The data in Table 5 show the association between the number of ISI-indexed publications and collaborative authorship; the question is whether collaboration has a direct or inverse relationship with publication number.
The average year-on-year change in collaborative papers was 40% during 1975-2005.There were 4693 collaborative papers amongst these publications, which rose to 462% by 2005, with an average yearon-year change of 60% -higher than the average of the total number of publications.The number of papers that were a product of domestic collaboration was 3139, which rose by 293% in 2005.The year-onyear average for internationally collaborative papers was 1824 in total, which registered a higher level of increase in percentile terms.The number grew 15-fold from 1975 to 2005, with an average year-on-year change of 243%.Although small in number, multi-country collaborations resulted in the production of 459 papers and showed the greatest growth.In 2005, multi-country collaboration reached 57 times that of the 1975 figure, with an average year-on-year change of 933%.Co-authored papers -including all three forms of domestic, international and multi-country collaborative papers -registered a higher level of growth than that for all papers.If co-authored papers only were compared with all publications, the year-on-year difference was over 50% (60% and 40%, respectively).This trend suggests that collaboration exerted an influence on the production of papers in the natural sciences.In other words, collaboration led to the generation of more papers than could be generated individually in the discipline in South Africa.
As there is evidence of association between collaboration and publication, the variables that affect collaboration were examined.Table 6 presents the Pearson's correlation coefficients of relevant variables.The degree of collaboration, measured on an additive scale that counted all the existing levels of internal-institutional, external-institutional, international and multi-country collaborations, seemingly correlates with the year of publication, the number of authors, the exclusivity of only South African authors in publications, South African authors within the same organisation, citation count, and the number of foreign countries involved in the production of publications.
The degree of collaboration was positively (and significantly) correlated with the year of publication, the number of citations and the number of other countries.The number of publications in local South African journals was negatively correlated with the degree of collaboration, which may positively affect the number of publications in overseas journals.Depending on the number of other countries involved in the output of publications, a positive change in the degree of collaboration can be envisaged.

Discussion
The growth trends in the production of scientific knowledge in the natural sciences in South Africa for the last three decades were similar to those in 'all subjects'.The steady increase in the average number of authors per publication in the natural sciences, juxtaposed with those for 'all subjects', conveys several points.Firstly, it denotes a declining tendency to produce knowledge individually (in sole-authored publications).Secondly, it explains the increasingly collaborative preferences of authors in the production of scientific knowledge.Thirdly, it denotes the increasing number of authors (from 2.24 to 5.11 authors per paper) working together in the production of research publications.Fourthly, it shows an ascending pattern in the participation of countries in internationally collaborative papers, which was not significantly different from that seen in 'all subjects' for South Africa.
Citation trends of the publications in the natural sciences did not provide evidence that the knowledge produced by South African scholars is being used by the academic community as much as the knowledge created in 'all subjects' in South Africa.In their analysis covering 30 countries, Glānzel and Lange 17 noted that the observed, expected, and relative citations of papers in biomedical research and chemistry produced higher rates of citations, more so in the case of papers that were the results of multi-country research projects.Another citation analysis of South African publications indicated that the citation count was also dependent on the type of collaboration that led to the production of the publication. 18e large majority of authors were from universities while the research institutes were the second major player in the production of publications.Pouris' 19 ranking of South African universities based on research output has also revealed the significant role of universities in research.Like South African researchers in the natural sciences, foreign partners originated mostly from universities and research institutes.The only difference between South African authors and their foreign partners affiliated to universities was that South African researchers had partners from government and industry, but, for foreign partners, government and industry partnerships were relatively negligible.This aspect of the leading role of universities in research has been reported in previous research. 4Pouris 3 observed that the higher education sector was responsible for 80% of the publication outputs of South Africans.What was striking in this analysis was the great difference between these two sectors (universities and research institutes) in the production of knowledge in the natural sciences.Research institutes produced only one-fifth of that of the universities, government less than one-tenth and industry only about 3% of the university sector.
Shrinking internal-institutional collaboration and expanding externalinstitutional collaboration point to significant trends in the disciplines in the natural sciences.South African researchers are looking outwardly rather than remaining within the limited confines of their own departments and institutions.This outlook was very clear in regard to international collaboration, which has in recent years exceeded domestic collaboration.More recently, international collaboration involving several countries has been expanding in the natural sciences.Mouton 20 found that inter-institutional and inter-sectoral collaboration are relatively low in South Africa.
Collaboration with foreign scientists is a feature that is remarkable in the natural sciences in South Africa.This finding corroborates that of other recent studies that reported an increase in international collaboration over national collaboration. 6Jacobs 6 also found that South Africans collaborate more frequently than their counterparts in the USA and the UK.
South Africans collaborate most with authors from developed countries in the global North -European and North American countries -while research alliances with countries in the global South are dismally few.Schubert and Sooryamoorthy 21 also noted this feature of South African collaboration.For Boshoff 22 , this scenario is an unequal and unbalanced partnership in North-South collaboration.Flemish researchers in the natural sciences also have shown similar interest in associating with international colleagues.Luwel 23 reported that about half of the total papers produced by Flemish natural science researchers in 1980-1996 were joint publications with foreign scientists.In Canadian research collaborations, as Larivière et al. 2 found, factors such as language and geographical proximity play an influential role.Jappe 24 also noted the influence of geographical locations on collaboration in earth and environmental sciences.A high level of collaboration, for a period of 23 years (1980-2002), was apparent in the natural sciences and engineering subjects in Canada. 2 Larivière et al. 2 reported that almost all papers were joint authored and the predominant form of collaboration was international rather than inter-institutional.Braun et al. 25 underlined the vigorous character of physics for international co-operation.As Davenport and Cronin 26 reported, joint authorships tend to vary according to disciplines and fields and the 'honorific' authorship phenomenon is common in the biomedical community.
Although causality cannot be established it is obvious from the data that the number of publications and collaboration are related.The likelihood of a publication produced through collaboration is more than that for the total number of publications.This finding suggests that increasing levels of collaboration -international over domestic -have an influential role in the production of publications in the natural sciences.This analysis has also highlighted the relationship between the degree of collaboration and some key variables including number of citations, the year of publication, publications in overseas journals, and the number of countries involved.
The above analysis and findings based on bibliographic records need to be viewed against the background of certain other factors that have influenced the production of publications in science in general and in the natural sciences in particular.
The analysis showed some trends in the production of publications.South African science was not free from the political environment.South Africa suffered a period of academic boycott on the activities conducted by South African researchers as a result of the apartheid policies of the government.International journals joined in this boycott and refused to accept South African contributions and avoided South African participants in conferences. 12This 'closed off' period from the mid-1980s to 1994 caused a decline in the publication productivity of South African scientists 5,6,14,11 ; publication was possible only in South African journals.Since 1994, there has been a positive change in the production of scientific publications, marking a favourable approach from the international community to collaborate with South African researchers. 14,15As noted before, changing funding policies also facilitated an increase in publications. 16Along with these, there has been an increase in the opportunities for scholars to publish their research, as new commercialised outlets have arisen.
Introduced in 1987, the funding system for universities made an impact on research and research output in South Africa. 10The funding formulae included a subsidy per research output in approved publications, and is still followed today with amendments.There were three formulae -the Holloway formula, the Van Wyk de Vries formula and the SAPSE (South African Post Secondary Education) formula -employed.The second and third formulae retained major components of the first and brought new features and parameters to suit the changing academic environment in South Africa.The SAPSE formula was later revised in the late 1980s and early 1990s. 27Scientists and academics were encouraged to publish in national and international journals.As early as the 1960s, scientific publications became the sole criterion for appointments and promotions, 28 which naturally increased the number of publications by South African scientists.
Drawing on the experiences of similar institutions elsewhere in the world and adopting their best features, the South African Council for Funding agencies that support research opt for inter-regional or international collaborative projects rather than individual projects located in a single institution or in a single country.Governmental agencies and private foundations formulate policies that aim at accelerating interinstitutional collaboration. 29 South African higher learning institutions, publication productivity is valued and an incentive system operates effectively to encourage it.By way of granting productivity units, the incentive system brings in research money to researchers.Since the introduction of a formula in 1984, research money is paid out to the university academics and scientists who publish papers in SAPSE-approved peer-reviewed journals that are on the list of the Department of Higher Education and Training.This process is the basis on which universities in South Africa obtain government funding.
There are already several policy initiatives that encourage collaboration. 30,31Quite clear in a number of policy documents is the emphasis on collaborative efforts within and outside the country.The Innovation Fund of the Department of Arts, Culture, Science and Technology has as its objective the advancement of transdisciplinary collaboration across sectors in South Africa. 32Centres of excellence, research teams, research centres and work groups are funded to facilitate and support collaborative efforts between disciplines, universities, industries, other institutions, regions and countries.As argued in a policy document, 33 South Africa has to open its doors widely for a strong and steady inflow of ideas and scientists.Also heard is the call for policies and programmes to promote unimpeded movement of scientific and technical information to the national and international systems and encourage South African scientists to participate in national, regional and international collaborative ventures. 31,34The existing networks within SADC (Southern African Development Community) and NEPAD (New Partnership for African Development) have turned out to be unsustainable for want of resources, 30 preventing effective collaboration with member countries.
Neglected in this discourse is the value and worth of the scientific data and knowledge that the developed countries acquire from the developing countries through such alliances.Quite evident from successful instances of collaboration is the balance in the collaboration configuration, such as when Africans provide access to local communities and non-Africans provide needed equipment and training. 35Collaborative activities function well when the partners recognise the reality that the source of funds is equivalent to the source of resources; none is superior to the other.It is true that for scientists in many developing countries (except for a few prestigious institutions within them), in contrast to those in developed countries, the cost of collaboration is at a premium.Basic essentials for conducting research -phone calls, postage, the Internet, email, stationery, printing and copying, library and database searches, assistance and local travel, not to mention equipment and laboratory material -are not always at the disposal of scientists in poorer countries.These are structural hurdles in materialising alliances, at least in the conceiving phase before funds are actually released.
The publication record of the natural sciences in South Africa for the last three decades does not indicate that natural science publications, collectively, compared with 'all subjects', have fared particularly well.
A real growth in these can, however, be achieved if appropriate and sustained measures from multiple sources are initiated.Producing young and new researchers is one such means to this end.

Table 1 :
Characteristics of publications in the natural sciences from SouthAfrica, 1975-2005

Figure 1 :
Figure 1: Growth of publications in the natural sciences and in 'all subjects' in South Africa, 1975-2005.

Figure 2 :
Figure 2: Year-on-year change in the production of publications in the natural sciences and in 'all subjects'.

Figure 3 :
Figure 3: Sectoral share of authors in the natural sciences, 1975-2005.The third major sector was government, which is comprised of departments, parks, museums and gardens.In 1975, this sector had a mean number of authors of 0.48, which by 2005 had reduced to 0.03.This decline was a loss of 94% of the original value reported in 1975 and was gradual until 1990, i.e. 65% in 1980, 29% in 1985, 31% in 1990 and 42% in 1990.After 2000 the fall was sudden.In 2000, authors representing the government sector consisted of just 8% of what the value had been in 1975 and declined further to 6% in 2005.Industries and hospitals did not make a significant contribution to the supply of authors in the natural sciences.
Scientific and Industrial Research (CSIR) was established.A major step in scientific research in the country, the CSIR paved the way for further growth in the crucial sectors of science and technology.It was an aim of the CSIR to provide both the government and industry with basic facilities of research through the creation of well-equipped national laboratories for fundamental and applied research across the country.The effects 10 Volume 109 | Number 7/8 July/August 2013 South African Journal of Science http://www.sajs.co.za of these initiatives were seen in the coming years in the production of scientific papers, including publications in the natural sciences.

Table 2 :
Co-authorship and collaboration in natural science publications from South Africa, 1975-

Table 3
presents the regional background of the overseas collaborators of South African researchers in the natural sciences.Collaborators were from all regions: Europe, North America, Australasia, Asia, Africa, the Middle East, Latin America and Eastern Europe.The order in terms of the mean number of representative regions was Europe, North America, Australasia, Asia, Eastern Europe, Latin America, the Middle East and Africa.The greatest number of partners was from European and North American countries.Australasia was represented in only about one-fourth of North America's share.Asia and Eastern Europe had about half the number of that of Australasia.Latin America contributed half of the Eastern Europe figure.The Middle East and Africa had the lowest representations.European scholars who worked with South African counterparts increased in number more than sevenfold: 1.5 times in 1995, 1.6 times in 2000 and 2.6 times in 2005.North American collaboration was not as high as European collaboration.Compared with the mean value of 0.241 for Europe it was 0.151 for North America.North American collaboration with South Africa expanded threefold of what it was in 1975, with the most remarkable expansion occurring in 1990 and thereafter.The average level of Asian collaboration for the entire period was below that of Australasian collaboration.This difference was balanced out in 2005 when Asian collaboration gained a slight edge over Australasia.Collaboration with other African countries was consistently insignificant.

Table 3 :
Regional and sectoral representation of foreign partners of South African authors in the natural sciences, 1975-2005 7 Volume 109 | Number 7/8 July/August 2013 South African Journal of Science http://www.sajs.co.za

Table 4 :
Countries of South African partners in natural sciences research, 1975-2005With regard to South African collaboration with the global North and the global South, it was evident that most collaborations involved countries in the North rather than in the South.Of the total collaboration, 86% was with scholars in European and North American countries.

Table 5 :
Collaborations and publications of scientists in the natural sciences in SouthAfrica, 1975-2005

Table 6 :
Correlation matrix of degree of collaboration in the natural sciences and other variables 9 Volume 109 | Number 7/8 July/August 2013 South African Journal of Science http://www.sajs.co.za