It is widely accepted that a requirement for honest handicap signallin
g is that higher-quality signallers pay lower marginal costs for adver
tising. This is a simple, powerful principle, but it is correct only i
f the fitness effects of fecundity and viability are strictly additive
. Additivity would not be expected from most life history models. The
general criterion for honest handicap signalling is that higher-qualit
y signallers must have higher marginal fitness effects of advertising.
This might result from higher benefits rather than lower costs. The g
eneral criterion implies the existence of a ridge on the fitness surfa
ce for two correlated characters, quality (or viability) and advertisi
ng. This has important implications for the design of experiments. Cri
tical tests of the handicap hypothesis should establish that signaller
s of different quality are on a rising fitness ridge because of differ
ent cost-benefit trade-offs. The further question of whether receivers
are maximizing their fitness requires additional experiments because
handicap signalling does not require that the receivers maximize their
fitness, only that they return benefits to signallers as an increasin
g function of the size of the signal. If receiver preferences are exag
gerated by sensory bias or indirect selection, the resulting exaggerat
ed signals may be consistent with the handicap principle. (C) 1998 The
Association fbr the Study of Animal Behaviour.