[Excerpt] This paper offers an historical dimension to the impact of trade unions on earnings by estimating the union wage effect in Britain in 1889-90 using data from the US Commissioner of Labor survey conducted at that time. The determinants of union status are also investigated in terms of a probit estimation using individual characteristics which may be correlated with union membership. The results of this first step are used in the computation of selectivity corrected estimates of the union wage effect. It is found that the effect of union membership on earnings at this time was of the order of 15%-20% and that this effect was similar at different skill levels. A broadly similar pattern is observed for industry groups, although the difference in the impact of unions on earnings across industries was greater than across skill groups.