1. Fourteen experiments with winter wheat on heavy-land farms from 1958–60 compared autumn dressings of sulphate of ammonia with equivalent early (March) or late (May) spring top-dressings. Dressings applied on two or all three dates of application were also compared.

2. Consistent and significant gains in yield from low (0·5 or 0·6 cwt. N/acre) and high (1·0 or 1·2 cwt. N/acre) levels were obtained at most centres, which were on fields under arable rotation.

3. With the low level of N, gains in yield were greatest from a single May top-dressing. There was no gain from dividing the dressing.

4. With the high level of N, dressings divided between autumn-May or March-May were best. March-May was the most consistently satisfactory treatment and gave the highest mean increases in yield in 1959 and 1960, but in 1958 autumn-May was better. Single May dressings gave slightly lower yields. There was no advantage from dividing N between all three dates of application.

5. Single applications of N in autumn gave lower yields than single applications in spring; supplementing autumn N with a late spring top-dressing gave almost the same yield as giving all the N in spring. Dividing the dressing in this way may be convenient on many heavy-land farms.

6. The percentage of N in the grain was highest with a late top-dressing of fertilizer-N, lowest with autumn N and intermediate with March top-dressings.