Adipocyte dimensions of subcutaneous fat, sampled by biopsy at the 13th rib, were measured in 20-week-old rams from lines of Texel-Oxford and Scottish Blackface sheep, both divergently selected for carcass lean content. A total of 163 animals were measured, with close to equal numbers per breed-line combination. In both breeds, the high (lean) selection line had significantly lower backfat depths (0·71 mm in the Texel-Oxford and 0·83 mm in the Scottish Blackface, s.e.d. = 0·14 and 0·13 mm, respectively), but body weight did not differ between the lines. The ultrasonic fat depth differences between the Texel-Oxford selected lines were accompanied by increases in adipocyte diameter, area, diameter standard deviation within each sample and implied cell number, calculated as the ratio of ultrasonic fat depth to average adipocyte diameter. In the Scottish Blackface sheep there were no selection line differences in adipocyte dimensions, but there was an increase in implied cell number in the line selected for increased fatness. Across breeds, ultrasonic fat depth was correlated with both adipocyte diameter and implied adipocyte number (r = 0·58 and 0·75, respectively), but these latter two measurements were uncorrelated with each other.

Scottish Blackface sheep have been selected for either increased or decreased predicted carcass lean content at 20 weeks of age, using an index combining ultrasonic backfat depth and body weight. The index was designed to alter carcass lean content without changing body weight. After 4 years there have been large responses in the index and fat depth, the proportional difference between the divergent lines in fat depth is 0·28, but body weight and ultrasonic muscle depth have not changed between the lines. The realized heritability for the index is 0·45. Heritabilities for the index, ultrasonic fat depth, body weight and ultrasonic muscle depth, calculated using multivariate restricted maximum likelihood, are 0·47 (s.e. = 0·14), 0·39 (s.e. 0·13), 0·23 (s.e. 0·12) and 0·36 (s.e. 0·14), respectively. The selected lines will be used to investigate relationships between carcass lean content, viability and maternal performance under the harsh environmental conditions encountered in the Scottish hill environment.