The evidence that earnings rise with firm size and that human capital affects earnings based on labour market data are two of the most robust empirical findings in economics. In contrast the evidence for scale economies in firm data is very weak. The limited direct evidence of human capital on firm productivity suggests that human capital is indeed productive and that the magnitudes are consistent with the findings based on individual data. The common objection to accepting the role of size and human capital as determinants of either earnings or productivity has been the role of unobserved factors. In this paper we investigate the roles of size and human capital in determining both earnings and productivity using a panel data set of matched labour firm data which allows us to control for such factors. We argue that neither the unobservable quality of labour, nor the unobservable characteristics of the workplace, is the source of the relationship between firm size and earnings, and that this effect can have a rent-sharing interpretation. For our data human capital is of minor importance in explaining either the distribution of earnings or productivity across firms of differing size.