1. Lucerne seed samples of known harvest history were treated by dry heat at various temperatures and durations; their germination capacity, emergence and establishment were tested by laboratory, greenhouse and field establishment tests on heavy Boil during 1952–6 at Cambridge.2. A successful and safe heat treatment was 42° C. for 5 days, increasing the germination by as much as 44% up to about 90% at the cost of hard seed.3. The establishment obtained on heavy soil from treated seed samples, although varying from season to season, was significantly higher (P > 0·001) at all seed rates than that obtained from untreated control samples.4. Significant differences were obtained in germination capacity and establishment between samples harvested at different conditions. High moisture content of seed and high percentage of dark brown seeds was associated with late harvest. Brown seeds had a significantly lower germination capacity than yellow seeds. Only samples with a high initial moisture and high percentage of brown seeds showed a substantial decrease in germination capacity during the storage period of 3 years.5. Although the mean number of plants established is significantly higher for higher seed rates the mean percentage of plants established decreases significantly with higher seed rates and the greatest wastage of seeds occurred at the high seed rate of 15 lb; per acre.6. The viability of seed and speed of germination and of emergence were the most important factors in determining the successful establishment of a crop on heavy soil.7. Hard seed is viable and could germinate and produce seedlings by the end of the season, but only a very small percentage of it could normally be established under field conditions. With untreated seed no significant correlations were established between laboratory germination and field establishment or between laboratory germination and greenhouse emergence, yet there were highly significant correlations in these cases for treated seed. With increased duration of laboratory germination test, corresponding to a decrease in the proportion of hard seed, values of r increased. Correlations between field establishment at three seed rates and greenhouse emergence were significant in the case of both untreated and treated seeds.8. When the hard seeds were added to germination or establishment the relationships between the results of these tests were very close and highly significant correlations were established for both lots of seed, untreated and treated. The high proportion of hard seeds was the main limiting factor preventing a close relation between tests for untreated seed.