Graduate Student Stipends: Greater reliance on the "traineeship" concept would facilitate a more purposeful and consistent treatment.

The situation of graduate students in physics is profoundly influenced by federal support, which has been applied with mixed, and occasionally contradictory, purpose. There are five important goals in the provision of graduate student stipends: (i) maintenance of "distinction for excellence"; (ii) rational distribution over scientific fields and subfields; (iii) constructive involvement of graduate students in undergraduate teaching; (iv) rational geographic and institutional distribution; and (v) administrative simplicity. These goals may be approached by making several adjustments of the present system. 1) The preeminence of individual predoctoral fellowships should be restored. They should be awarded sparingly to at most five percent of the graduate student population. 2) The great majority of stipends should take the form of traineeships. They should be assigned to departments according to rational criteria. 3) Research assistantships should be continued at a reduced level. Their maintenance will protect research activities of specific public interest when necessary. 4) Teaching should be incorporated into the normal responsibilities of fellows, trainees, and research assistants. Teaching assistance should be regarded as a budgeting category and not as an identifying characteristic of a particular group of students.