In 14 experiments made from 1968 to 1970 yields of, percentages of N in and amounts of N removed by the grain were compared: (1) when N was injected, either as aqueous ammonia (with or without a combine-drilled ‘starter’ dose of N) or as aqueous urea, or when broadcast as ammonium nitrate (‘Nitro-Chalk 21’) and: (2) when comparable liquid (14–6–8) and granular (20–10–10) NPK compound fertilizers were applied, either over the seed bed (sprayed or broadcast), or combine-drilled. All were applied to give 63 or 126kgN/ha.

Barley yields were increased significantly more by combine-drilled than by broadcast PK (0–20–20) fertilizer, especially when only 32 kg P2O5 and K2O/ha was given. N was better combine-drilled than broadcast when 63 kg/ha was given, but not with 126 kg/ha. Soil analyses suggested a limiting value for P.

Yields and N recoveries were largest when one-fifth of the N was combine-drilled (as 6–15–15) and the remainder injected as aqueous ammonia in bands 30 cm apart and 12 cm deep. Injecting all the N, either as aqueous ammonia or as urea, gave yields a little smaller than this, but still larger than from broadcasting ammonium nitrate.

The liquid NPK fertilizer gave smaller yields than the granular one in the two dry springs (1968 and 1969), but similar yields in the cool wet one (1970), both when applied over the seed bed and when combine-drilled. The smaller yields with the liquid fertilizer were associated with a severe check and loss of plant when it was combinedrilled (presumably caused by urea) and poorer growth after it was sprayed (presumably through loss of N to the air and greater dispersal of P in the soil). The single amount of each fertilizer increased yields more when combine-drilled than when broadcast, but not the double amount, which, especially as liquid, was safer when surface-applied. Effects of site and season upon mean yields were large.