I would like to emphasize the diverse strategies and instruments that should be employed in future studies of extragalactic H II regions. Large-aperture, ground-based telescopes will play a key role in obtaining measurements of weak lines (e.g. [O III] 4363Å) in distant, low-luminosity, and high-metallicity nebulae, enabling more reliable "direct" abundance determinations to be made. A good deal more can also be done with such telescopes in studying spatial inhomogeneities (Rosa 1983). Naturally, the Hubble Space Telescope will also be a prime tool for studying extragalactic H II regions. Not only will it yield carbon abundances for many more objects, but the high spatial resolution obtainable with HST will allow better isolation of the nebular spectrum from the stellar continuum, as well as yielding a great deal of information about the properties of the ionizing stars.
Other tools besides optical/UV telescopes should not be overlooked. In particular, the investigation of infrared emission lines from extragalactic H II regions is likely to prove rewarding (e.g. Dinerstein 1986). As of now the surface has barely been scratched, especially for the far-infrared lines; the current largest far-infrared telescope, NASA's KAO, can barely reach the very nearby H II regions in M 82 (Duffy et al. 1987) and M 33 (Dinerstein et al. 1987). Once instruments such as SOFIA (a 3 m-class airborne telescope) and SIRTF (the Space Infrared Telescope Facility) are in operation, many of the regions which have been the subjects of intense study in the optical will become accessible in the infrared. Finally, one should not overlook the potential contribution of fast contemporary computers in making it practical to undertake realistic calculations of stellar model atmospheres and stellar evolutionary tracks which include non-LTE and metallicity effects and stellar winds (e.g. Oloffson 1989). The prospects seem excellent for answering some of the unsettled questions listed above, in the near future.
Acknowledgments. I would like to thank the meeting organizers for providing support which enabled me to participate in this conference, and Don Garnett and Evan Skillman for many helpful discussions during the preparation of this paper.