The results were less clear, however, concerning the relative importance in writingof fact and interpretation. By this point in the course both ‘arts’ and ‘science’ studentsseemed to be aware that essays in the history of science involved the evaluation ofdifferent views. ‘Arts’ students, however, more often indicated that they had expecteda more factual course, preferred writing about facts than opinions and considered‘getting the facts down’ as the most important criterion. This orientation towardsfacts in the ‘arts’ students’ questionnaire responses appears surprising, given that intheir writing they were more likely than ‘science’ students to represent ideas as provi-sional and mediated rather than factual.This apparent discrepancy may perhaps be resolved by distinguishing betweenstudents’ perception of an issue as problematic and their actual performance withrespect to that issue. ‘Arts’ students’ views of the role of fact and interpretation mayindicate an awareness that the representation of reality is not straightforward and thatessays require a sophisticated discussion and evaluation of different points of view.They may deal more effectively with these issues precisely because they realise thatthey are problematic. Interviews with ‘science’ students suggested that they were notalways conscious of the significance of this type of discussion and might perceive it as‘waffle’ or ‘padding’.

Kuhn (1970, p. 167) commented that science education tends to elide the processthrough which knowledge has been constructed, whereas students of other subjectsare exposed to varying interpretations over time. As a result, he suggested, sciencestudents are blind to the history of their subject, seeing it only as unproblematicprogress. The interview data suggest that this is indeed a point of difference betweenthe ‘arts’ and ‘science’ students in this sample. While both of them tend to have adualistic view of science itself, the ‘arts’ students seem to be more at ease with arelativistic view of knowledge in history.

Seven of the ‘arts’ students described a process of this sort, compared with only twoof the ‘science’ students. There was, however, another approach to revision, involvingonly one revision cycle. This was mentioned by five interviewees, four of whom werefrom ‘science’ backgrounds. Um ... rewriting? No. I can probably, once I’ve got the, I’ve got the feel of it, it probablytakes me a couple of hours to write, and then, shuffling stuff around, ... it’ll probably takeme, I don’t know, a morning or something to do a fair draft of it. (Ewan, 2002, science)Only one ‘arts’ student mentioned using a single revision cycle, and he had originallygraduated in science before starting his OU arts study

Although some ‘science’ students reported similar problems, it was only ‘science’students who talked in terms of ‘padding out’ their answers in order to reach therequired length: I’m more this, get all the facts down, yes it’s only three hundred words, but that’s it in anutshell. And it’s a lot harder then to flower it up to say either five hundred words or athousand words. (Larry, 2002, science)I’m not used to waffling I think that’s the problem. A lot of the art students say oh I’vewritten too much, ... and I have the opposite problem I kind of write down what Ithink the answer’s and I’ve only got like 200 words and I have to pad it out. (Ruth,2003, science)The tendency for some ‘science’ students to write relatively short essays may berelated to their conceptions of knowledge. If it is seen as factual, then once the factshave been stated, the student might see the task as complete; as Larry said, ‘that’s itin a nutshell’. If knowledge is relativistic, however, then competing views are equallyworthy of consideration and greater elaboration is needed to make a case

While the ‘arts’ students frequently described a strug-gle to make their essay ‘flow’, the ‘science’ students did not talk about textual struc-ture as problematic

Geisler (1994) and Russell and Yañez (2002) discuss a comparable situation in theUSA, where to fulfil general education requirements, undergraduates take a numberof disciplinary courses in fields which are not their major. They note the contradic-tions involved in conflating the aims of general education and disciplinary encultura-tion, with lecturers using a disciplinary discourse that is not only unfamiliar tostudents, but also seen as irrelevant to their individual aims and aspirations. Similarly,Moore (2000) discusses the tension between integration and disciplinarity in an inter-disciplinary foundation course in South Africa, voicing concerns that the attempt topromote generic competences risks undermining the disciplinary basis of academicperformance (p. 192).

The distinction between hard and soft fieldsrelates to the extent to which knowledge is constructed on the basis of a frameworkof shared assumptions. The pure sciences (hard) typically maintain a degree ofinternal unity over aims, methods of investigation and evaluation criteria, which maycome to be seen as derived from reality itself, rather than constructed by disciplinaryconvention. The humanities and social sciences (soft), in contrast, tend to becharacterized by internal discord, encouraging a view of knowledge as a matter ofinterpretation.

‘However’ is a textual theme with the function of indicating the relationship of theclause to the preceding text; ‘it is apparent’ is an interpersonal theme with the func-tion of indicating the writer’s stance towards the proposition that follows; ‘during thesecond half of the sixteenth century’ is an experiential theme providing informationabout circumstances surrounding the event or situation. In the discussion that followsI refer to these three types of non-subject theme as orienting themes. Unlike thesubject, none of them is grammatically compulsory and their use reflects a choicemade by the writer about how to frame the proposition presented within the clausecomplex.These orienting themes were consistently more common in the ‘arts’ students’essays, and the difference between the two groups was highly significant (t= 2.865,p < 0.006). ‘Arts’ students used on average 31.50 textual and 15.14 interpersonalelements in every 100 clause complexes, compared to 24.28 textual and 9.75 inter-personal elements for the ‘science’ group. They also tended to use more clausecomplexes containing an experiential orienting theme, although this difference wasnot significant. Since essays which used more orienting themes were also significantlymore likely to receive a higher mark (t= 2.336, p< 0.023), it is clearly worth investi-gating further the differing ways in which these were deployed by ‘arts’ and ‘science’students.

Research suggests that students majoring in hard fields with a high degree ofdisciplinary consensus are more likely to subscribe to beliefs in absolute knowledgethan those majoring in soft fields, and that these beliefs may be encouraged byaspects of the disciplinary context in which they work (Paulsen & Wells, 1998;Schommer-Aikins et al., 2003). Neumann (2001) reviews evidence of disciplinarydifferences in a number of aspects of teaching and learning, noting that soft disci-plines tend to emphasize critical thinking, oral and written expression, and analysisand synthesis of course content, while hard disciplines tend to emphasize skills indealing with facts and figures, with little writing required beyond the exposition ofexperimental results. In a large-scale undergraduate survey Entwistle and Tait(1995) found that students’ learning styles varied between different disciplines in linewith the demands of their course. Students of science and economics, for example,were more likely to use surface strategies, perhaps encouraged by assessmentpatterns that emphasized the reproduction of facts. In contrast, markers in historyand English were likely to penalize a reproducing orientation and a serialist (listing)style (Entwistle & Tait, 1995, p. 96).

Such tutor comments suggest that ‘science’ students are less ready to criticallyevaluate source material, a feature that can be related to the tendency already notedin their writing to downplay the role of human interpretation in the construction ofknowledge