What is the role of a PhD supervisor? Some students receive ample help and training while others are abandoned. John Davies reports Mark Stevenson is a PhD student at a well-known institute of ÖйúAƬ. He has a good first degree and practical experience in his chosen field. But he does not expect to get his doctorate. The reason: a "total breakdown" in his relationship with his supervisor.

"I was taken on as a research assistant without meeting (him), and when I arrived it turned out he didn't have a PhD and hadn't supervised before," says Stevenson. "There has been a breakdown in our relationship. But my funding depends on my supervisor - he's an expert in the area - and the institution doesn't really have anyone to replace him. I'm an outsider and he is an insider - anything I say carries no weight. I do have a second supervisor, but she doesn't have the time to see me.

"There have been four postgraduates in the past two years in my department, and I'm the only one left. Maybe if I'm lucky I'll get an MPhil here, but I don't have any control over my funding, and I'll need a reference from this institution if I look for a job."

Stevenson's experience may not be typical, but he is by no means exceptional. Tricia Skuse, who is finishing a psychology PhD, has observed "a wide range of different supervisory problems" through an organisation she helped set up, the Network for Postgraduate Research into Adolescence. The network, now more than three years old, was set up "to reduce both academic and personal isolation among PhD students . . . involved in some aspect of adolescence in a range of disciplines including psychology, education, social work and health studies".

ÖйúAƬ

"I think one of the fundamental problems with PhDs in Britain is that nothing's standardised," says Skuse. "I have friends who start their PhD, see a supervisor a few times and then they're left to fend for themselves. In other universities supervisors will go with you and help you set up your fieldwork, or help you do your analysis and give you a training in research skills."

Carl Green is definitely in the fending-for-himself category. An American over here for a humanities PhD at an "old" redbrick university, he observes that his tuition fee - he is self-funded - "bought me a library card but little else". He used to see his supervisor about once every two months; now the supervisor is elsewhere for a year and he does not see him at all. As a result, Green has "opted out" and is taking a taught course.

ÖйúAƬ

"My initial problem was that my expectations were uninformed," he says. "In the United States, doctoral programmes (in humanities) begin with one or two years of structured seminars close to your research interests. That gives you the background expertise from which you narrow down. Coming from the US, I expected some guidance, instead of just being dumped in a library with half a million volumes." Somewhere near the opposite end of the spectrum is Paul Hobson. In his first year of a geography PhD at the University College of Swansea, he has "at least one weekly meeting of two hours" with his supervisor. "I can get access to him quite easily, and I'm not his only doctoral student," he says.

What, then, makes for a good student-supervisor relationship? Turn to Research Student and Supervisor: An approach to Good Supervisory Practice, a booklet published in 1992 by the predecessor of the Engineering and Physical Sciences Research Council. You will find a 20-item checklist of "practices which we think, if generally adopted, would lead to many more students completing their PhD in three years". Its questions include: "Is there a departmental document available to students and supervisors laying out the department's view of good practice?" and "What steps are taken to try and make a good match between a supervisor and the prospective student?" Other questions, for the supervisor, include: "Does the student get a mock viva between six and 12 months before the submission date?" and for the student: "Are your records in good order and could you answer a question on something you did six months ago?"

A research council such as the EPSRC is not alone in looking for codes of practice. The ÖйúAƬ Quality Council's soon-to-be-published Learning from Audit 2 will contain a chapter on postgraduate research degree students that notes the variations in supervision practice. As a result, HEQC assistant director Chris Haslam is seeking "to bring together the various codes of practice that have been issued over the years into a composite framework". A working party on the subject will, he hopes, be able to report by the late spring or early summer.

Among the institutions commended in Learning from Audit 2 is the London School of Economics. There, every year, all PhD students and supervisors have to fill in a progress report form. The students also have to complete a confidential questionnaire that is only seen by the dean of LSE's graduate school. The LSE says he "uses the information as he sees fit".

"We have codes of practice which make explicit what to do when things go wrong," says assistant registrar Catherine Manthorpe. "People don't always follow the procedures and rules, but we try to have fall-back mechanisms and safety nets."

The newly-established University of Luton has awarded very few PhDs as yet, although it has about 130 students enrolled for research degrees. Luton was not created out of a polytechnic, in the wave of conversions made several years ago. Because of this it was only granted research degree-awarding powers after an HEQC-led scrutiny - a process that ensured that the then Luton College of ÖйúAƬ had to have training programmes for supervisors, research methods courses for potential PhD students, and a tight monitoring system in place.

Stephen Fallows, reader in educational development at Luton, says: "If you take an average group of PhD supervisors in an older university and ask them to go through the procedures that we had to, I wonder how they would fare?" In some older universities, he goes on, "in theory you can finish a PhD one week and start supervising one on your own the week after - something that couldn't happen here".

He also contrasts his Luton experience with his previous university, Bradford, where he taught and supervised PhD students. "I never had any training there. There was no handbook that said 'This is good practice'. I learned on the job. But I was lucky to work alongside experienced colleagues who showed me the ropes." Luton has a team supervision system; each PhD student has three supervisors, of which at least one must have experience of supervision through to successful completion.

ÖйúAƬ

Bradford now has a more formal approach, says Nick Buck, senior assistant registrar. "We have had guidance notes for students and supervisors since 1993, and there's now an enlarged booklet for supervisors and another one for students. First-time supervisors can't supervise on their own, and there's a procedure if students have difficulty with their supervision arrangements."

ÖйúAƬ

But rules on paper are not necessarily rules in practice. John Hockey, who has written about doctoral education and is researching at Cheltenham and Gloucester College of ÖйúAƬ, says: "You can put things down on paper, but unless you hold people to that contract it's so much hot air."

The HEQC's Haslam says: "Although codes of practice have been introduced, the evidence that they are monitored is variable."

So, although it is good if a university has its own internal procedures for resolving supervisor-student problems, if a resolution is not possible, to whom can students appeal externally?

Research councils, if they are the source of a student's funding, may help. Michael Jubb, secretary to the British Academy's humanities research board says: "Students do come to us and we do take up issues with them and their supervisors. But we regard it as primarily the responsibility of the individual institution. We can't police every award."

His board does, however, have a working group looking at codes of practice and "the expectations we have of students with awards in the humanities". It is due to finish its deliberations this summer.

And then there is the HEQC. Under the terms of the Citizen's Charter, it will deal with complaints from students if they have been given misleading information and if they have already gone through internal procedures without satisfaction.

Bridget Rogers, head of communications at the quality council, says an individual case has occasionally resulted in the establishment of a complaints procedure at the relevant university. "But we have no authority or control over universities . . . We're toothless," she says. With the number of complaints going up - "mostly about academic matters, poor supervision, the usual things" - Rogers is "very sorry" more cannot be done. "There is no ombudsman for ÖйúAƬ. I feel there should be a proper procedure - it's very unsatisfactory.

"I had 20 years experience at the Council for National Academic Awards (the former polytechnic degrees awarding body), and there we had large fangs. When a complaint came to us we required the institution to put it right. Now it's a different ball game."

Perhaps when it reports, the national committee on postgraduate education headed by Manchester University vice chancellor Martin Harris will have recommendations to make on supervision.

The LSE's code of practice talks of a "harmonious relationship" between research student and supervisor that "can only be established and maintained if both participants understand each others' concerns, (and) treat each other with courtesy". But can courtesy be legislated for? How can guidelines help when, in Mark Stevenson's words, "it comes down to two personalities who can't get on".

As Tricia Skuse observes: "I think part of the process of doing a PhD is that you may have problems with your supervisors. It's part of your growing independence and autonomy."

ÖйúAƬ

Mark Stevenson and Carl Green are false names.