Work by social constructionists over the past decade and a half has reenforced the epistemological pessimist's despair that our system of science could ever be a mirror of nature. Realists argue that the amazing success of modern science at precise prediction and control indicates just the contrary. In response, social constructionists often point out that these successes seldom apply to the world as it comes naturally, but only as it is reconstructed in the scientist's laboratory. But this does not explain how scientific accounts, which are as wide of the mark as constructionists suppose in their general descriptions of the natural world, could be so effective in a limited environment, even if that environment is as benignly constructed as possible. This paper suggests that a more wholistic view of nature than modern science usually presupposes may provide an answer.