The main problem in writing letters of recommendation derives from a basic contradiction: the recommender wants to promote the candidate, yet at the same time he or she needs to convey the impression of giving an objective evaluation. I see no way around this problem. Unnuanced encomium will inspire disbelief, and unadorned frankness will be self-defeating. The most common strategy is to begin the recommendation with a barrage of praise and then to add nuances that can sound somewhat critical. On the whole, this works: the recipient is assumed to be savvy enough to discount for the rhetoric while understanding that the recommended is a less-than-perfect human being, like the rest of us. The trick is to get the balance right.

Compounding this fundamental problem is the general inflation that has hit recommendations just as it has undermined grades, the currency, and so much else in our culture. As the saying goes, even nostalgia isn't what it used to be. Time was when recommendations from the U.K. were especially effective, because they were tough and believable. But the English have succumbed to the American disease, and recommendations from the Continent tend to be worthless. (French professors, for example, often restrict their recommendations to a few sentences, which read like an order to promote the candidate in deference to the recommender's eminence.) How to cope with this general tendency? If you rein in your rhetoric, you may punish your student. If you give it its head, you may destroy your credibility. Here, too, I can recommend only a middle way with a cautionary note: bad money chases out good.

These difficulties are mitigated by circumstances, for there are recommendations and recommendations. The tone and style of the letter you write will depend on circumstances, notably the situation of the person you are recommending—an undergraduate, a graduate student, a professor—and your own position: if you are a 60-year-old professor at the Collège de France, you may be able to afford more frankness than a young historian in a community college. On the whole, recommendations for graduating seniors can include fewer negative comments and can stress personal qualities such as maturity and the ability to perform tasks effectively. Those for graduate students should be more detailed, with references to specific papers and performance in seminars. Those for colleagues should include honest comments about your relations: "Having gone to graduate school with X and followed her work for the past 20 years, I admit to some bias but think I can offer a well-informed evaluation." A recommendation for a job will differ from a recommendation for a summer fellowship or some other, less crucial distinction.

In general, the more urgent and serious the occasion, the greater the need for detail and thoroughness. Recommendations for graduate students demand special care. They should include specific references to scholarly work and enough description of the candidate's personal qualities to make him or her stand out as an individual. Owing to the general inflation, an informal code has evolved. Phrases such as "diligent" or "hard-working" tend to mean "mediocre" or "pedestrian." You should be careful in using them, and you can reinforce the credibility of your remarks by comparing the candidate to others or indicating his or her relative ability: "among the best students I have taught" or "one of the top three students I have encountered during the last decade." Many universities keep placement folders for their graduate students. In that case, you may prefer to write one, all-purpose letter and to add specific letters as occasions arise. Of course, job recommendations should include an assessment of the candidate's scholarship. On the whole, it is better to be relatively brief. The recipients of the letter will have received examples of the work and will be able to make their own judgments. But it is important to indicate the originality and importance of the scholarship by putting it in the context of other work.

Finally, many of us receive requests for recommendations from people we hardly know or do not know at all. I think it legitimate to refuse them. What to do with the many requests that come from foundations and government agencies who bombard you with requests for evaluations of applications for grants? We need to balance our obligations to scholarship in general against our need to get on with our own work and to have some time left over to see our families or to enjoy life. Some of us could spend the entire year doing nothing but recommendations. Everyone will have to find his or her own balance. Some may set a limit. Some may favor requests that bear on their own area of expertise. Some may restrict themselves to their own discipline or to requests that come from their own country. The quantity and variety of requests make it impossible to suggest a general rule, but I think the greatest danger may come from the super-ego. True, most of us have benefited from the help of those who recommended us and we have a duty to help others, but we can reach a point of satiety. We have a right to say no.

—Robert Darnton is professor of history at Princeton University.