figures). A number of restaurant chains in the Toronto area report decreased sales during the summer—not only fewer customers, as might be expected because of vacations, but also a decline in the average purchase per customer (range of decrease: 2 to 20 percent). The only exceptions to this rule were a chain of restaurants specializing in salads and two chains specializing in ice cream desserts.

Fever, of course, is not an exogenous source of heat, but it may be considered as a means of inducing heat in the internal environment. Fever is usually associated with decreased appetite, which follows from most analyses of feeding in which thermoregulation is a consideration. Raising the internal temperature ought to trigger compensatory decreases in feeding if feeding threatens to raise the internal temperature even further. This conclusion is premised on the notion that fever represents a state of hyperthermia relative to some internal optimum. Note that it does not follow that if hyperthermia suppresses voluntary appetite, then voluntary appetite suppression necessarily indicates hyperthermia. Appetite suppression and indications of hypothermia coexist, for instance, in anorexia nervosa (Garfinkel and Gamer, 1982).

The possibility remains, of course, that fever might not represent true hyperthermia but rather a resetting of the thermoregulatory set-point at a higher level (Mrosovsky, 1990); in this case, the body might "want" to maintain a higher temperature, and a decline in feeding would not be expected. Intraperitoneal injection of interleukin-1, normally released in the presence of pathogens, raises body temperature and ordinarily is associated with appetite suppression; but when injected intracerebroventricularly, interleukin-1 raises body temperature in rats without affecting intake (McCarthy et al., 1986). This suggests that fever-induced anorexia may not be mediated by thermic mechanisms. Conversely, injecting endotoxin substantially lowers intake even when temperature elevation is prevented by administration of sodium salicylate (Baile et al., 1981; McCarthy et al., 1984), although the suppression may be less than when fever is not prevented (Baile et al., 1981). Another study, in which rats' body weights were lowered before the administration of interleukin-1, found that despite elevated body temperature, the animals were initially hyperphagic in defense of an albeit subnormal body weight (Mrosovsky et al., 1989); in other words, it may be that pathogens—or the interleukin-1 stimulated by them—produce a lowered BW set-point, pulling appetite down, independent of heat compensation (see also O'Reilly et al., 1989). There are plausible adaptive reasons why maintaining a lowered BW might help to fight or "starve" pathogens (Murray and Murray, 1977).