essarily from some place, and the relevant scale must be specified. These general conclusions follow from relativity theory (Prigogene and Stengers, 1984). But ecological processes, especially, can also be described on countless alternative scales (Levin, 1992). How do we know the scale on which to model an environmental "problem" and measure our progress toward solving it? Pure ecology cannot be the only guide because the descriptions of nature it provides are too numerous and also incommensurate because of the differing scales they embody. An important goal of environmental ethicists and policy analysts should therefore be to ascertain which natural dynamics are associated with important social values. What is needed is a more encompassing, interdisciplinary discussion of environmental values and goals. Of course such a discussion must be based on the best science, but establishing crucial links between ecological processes and environmental policy can be understood only in conjunction with a process of articulating important social values. In the process, the boundary between theoretical ecology and applied disciplines, such as restoration ecology and ecological engineering, will no doubt be blurred.
Determining which dynamics require special protection is an evaluative task that can only be done from a perspective; and identifying scales and bifurcation points is on this view a crucial aspect of management. All of this is part of defining what is a healthy system and when a system maintains its integrity.1 Scale is at the heart of all these problems, but we cannot choose appropriate scales to focus on until we understand both how ecological functions and processes work at various levels and how these functions and processes are associated with social values. Fully understanding this point will require that we reconsider the assumptions of "value-neutral science." In the modern, Cartesian-Newtonian period, and especially in the positivist era of science since 1900, description and evaluation have been regarded as separable steps in the process of understanding and acting in nature. It has been thought that natural systems and their products can first be described and that evaluations can then be applied to these "objective" descriptions as a separable step in the process of judgment and choice of actions. What we know now is that our choice of descriptive concepts shapes our perceptions (Kuhn, 1962; Quine, 1960). Even more important, conceptual and theoretical choices are colored by our values because we choose our theories in the process of accomplishing conscious or unconscious goals through action.
The recognition that some of our concepts embody both factual and evaluative content is really only a special case of a much more general phenomenon— description and prescription are so entwined in our use of language that it is often impossible to separate them in ordinary discourse (Nelson, 1995; Williams, 1985). This interpenetration of values and facts in ordinary discourse can be cited as one of the reasons scientific disciplines create more precise vocabularies; but, however useful an introduced special vocabulary, such formalistic languages cannot ultimately achieve pure description and at the same time be rich enough to guide