From Stephen Jay Gould, Wonderful Life, 1989.  New York: Norton.  pp 139-140

Universities operate one of the few survivors of the old apprenticeship system in their programs for awarding doctoral degrees. Consider the anomaly. You spend your entire educational career, from kindergarten to college, becoming more and more independent of the power of individual teachers (cross your first-grade teacher and your life can be hell for a year; displease a college professor, and the worst you can do is fail a single course). Then you become an adult, and you decide to continue for a Ph.D. So what do you do? You find a person whose research intrigues you, and sign on (if he will accept and support you) as a part of a team.

In some fields, particularly those with large and expensive laboratories dedicated to the solution of definite problems, you must abandon all thought of independence, and work upon an assigned topic for a dissertation (choice in research is a luxury of later postdoctoral appointments). In more genial and individualistic fields like paleontology, you are usually given fair latitude in choosing a topic, and may emerge with a topic uniquely your own. But in any case you are an apprentice, and you are under your mentor's thumb--more securely than at any time since the early years of primary school. If you and he have a falling out, you quit, or pack up and go elsewhere. If you work well together, and your mentor's ties to the profession are secure, you will get your degree and, by virtue of his influence and your proven accomplishments, your first decent job.

It's a strange system with much to criticize, but it works in its own odd way. At some point, you just can't proceed any further with courses and books; you have to hang around with someone who is doing research well. (And you need to be on hand, and ready to assimilate, all the time, every day; you can't just show up on Thursday afternoon at two for a lesson in separating parts from counterparts). The system does produce its horrors--exploitive professors who divert the flow of youthful brilliance and enthusiasm into their own dry wells, and provide nothing in return. But when it works (as it does rather more often than a cynic might expect, given the lack of checks and balances), I cannot imagine a better training.

Many students don't understand the system. They apply to a school because it has a general reputation or resides in a city they like. Wrong, dead wrong. You apply to work with a particular person. As in the old apprenticeship system of guilds, mentor and student are bound by mutual obligations; this is no one-way street. Mentors must, above all, find and provide financial support for their students. (Intellectual guidance is, of course, more fundamental, but this part of the game is a pleasure. The real crunch is the search for funding. Many leading professors spend at least half their time raising grant support for students). What do mentors get in return? This reciprocation is more subtle, and often not understood outside our guild. The answer, strange as this may sound, is fealty in the genealogical sense.

The work of graduate students is part of a mentor's reputation forever, because we trace intellectual lineages in this manner. I was Norman Newell's student, and everything that I ever do, as long as I live, will be read as his legacy (and, if I screw up, will redound to his detriment--though not so seriously, for we recognize a necessary asymmetry: errors are personal, successes part of the lineage). I happily accept this tradition and swear allegiance to it--and not for motives of abstract approbation but because, again as with the old apprenticeship system, I get my turn to profit in the next generation. As my greatest joy in twenty years at Harvard, I have been blessed with several truly brilliant students. The greatest benefit is an exciting lab atmosphere for the moment--but I am not insensible to the custom that their future successes shall be read, in however small a part, as mine also.