The range of both previous and current published values for the expansion rate, or Hubble constant, H0 (see Figure 1a), attest to the difficulty of measuring this parameter accurately. Fortunately, the past 15 years has seen a series of substantive improvements leading toward the measurement of a more accurate value of H0. Indeed, it is quite likely that the 1- uncertainty in H0 is now approaching 10%, a significant advance over the factor-of-two uncertainty that lingered for decades. Briefly, the significant progress can be mainly attributed to the replacement of photographic cameras (used in this context from the 1920's to the 1980's) by solid-state detectors, as well as to both the development of several completely new, and the refinement of existing, methods for measuring extragalactic distances and H0 (e.g., Livio, Donahue & Panagia 1997; Freedman 1997b).
Currently there are many empirical routes to the determination of H0; these fall into the following completely independent and very broad categories: 1) the gravitational lens time delay method, 2) the Sunyaev-Zel'dovich method for clusters, and 3) the extragalactic distance scale. In the latter category, there are several independent methods for measuring distances on the largest scales (including supernovae), but most of these methods share common, empirical calibrations at their base. In the future, another independent determination of H0, from measurements of anisotropies in the cosmic microwave background, may also be feasible, if the physical basis for the anisotropies can be well-established.
Each of the above methods carries its own susceptibility to systematic errors, but the methods as listed here, have completely independent systematics. If history in this field has taught us nothing else, it offers the following important message: systematic errors have dominated, and continue to dominate, the measurement of H0. It is therefore vital to measure H0 using a variety of methods, and to test for the systematics that are affecting each of the different kinds of techniques.
Not all of these methods have yet been tested to the same degree. Important progress is being made on all fronts; however, some methods are still limited by sample size and small-number statistics. For example, method 1), the gravitational time delay method, has only two well-studied lens systems to date: 0957+561 and PG 1115. The great advantage of both methods 1) and 2), however, is that they measure H0 at very large distances, independent of the need for any local calibration.