Decays of sigma, kappa, a0(980) and fo(980)

Ratios of coupling constants for these decays are compared with q-qbar predictions and Jaffe's 4-quark model. In both models, the predicted ratio g^2(kappa -> K-pi)/ g^2(sigma -> pi-pi) is much too small. Also, for q-qbar, the predicted ratio g^2(kappa -> K-eta ')/g^2(kappa -> K-pi) is much larger than observed. Both models fail for g^2(fo -> KK)/g^2(a0 -> KK). This ratio requires that fo has a dominant KK component. It arises naturally because the fo pole lies very close to the KK threshold, giving its wave function a long KK tail.