Results from a long term study of wheat experiments on Rothamsted farm were examined by multiple regression analysis to estimate the size of the contributions made by four weather factors to the year-to-year variance of grain protein.Whether considered singly or in either of two chosen models, air temperature during the early part of the grain-filling period had a positive effect, incident radiation during the early or middle part a smaller negative effect. Both models accounted for 25% of the total variance or 60% of the year-to-year variance. After fitting either model winter leaching and maximum potential soil-moisture deficit were not significant. Closely matching results were obtained by including 15 nearby farms. Some soil factors were investigated in a similar manner. A soil N index accounted for 30% of the total variance, with grainN% of 1.95 corresponding to Index 0 and 2.23 to the average of 1 and 2. In models combining the best weather and best soil variables (N Index and wetness class), the variance accounted for was 46%, with all effects significant. There were significant interactions of radiation, winter leaching and soil wetness class with the response to fertiliser nitrogen. As the yield data are less well based than those for grainN% their relationships were examined only in parallel to the main study.

At two English forest nurseries, transplants of five conifer species-Picea sitchensis, Picea abies, Tsuga heterophylla, Abies grandis, and Pinus contorta-were grown with fertilizer supplying N, P, K, and Mg in amounts intended to be adequate for producing healthy green trees with nutrient concentrations in the 'sufficiency range' as determined by earlier experiments. 'Luxury uptake' of nitrogen was obtained with top-dressings of 'Nitro-Chalk' applied in the nursery during early September, when top growth had nearly ceased. Tests of the effect of this extra N on forest establishment were repeated in four successive years under a wide range of soil and climatic conditions, keeping the trees in a cold store during each winter and planting them on forest sites in England, Scotland, or Wales during the following spring. Except for Grand fir, nitrogen advanced bud-break of all species during the first summer after planting and had no deleterious influence on survival. It tended to increase growth of Sitka spruce during the season after planting, but in later years the differences became small in relation to tree size. The effects on other species were small, except for one considerable decrease in the growth response of Grand fir at a single site.Frost damage of Sitka spruce of Washington origin was severe on a Welsh and a Scottish site where this frost-sensitive provenance would not normally be grown. At the Welsh, but not the Scottish site, the nitrogen treatment increased the damage.In the few experiments (confined to Picea sitchensis) which tested late-season potassium in the nursery, K concentrations were increased from deficiency to barely sufficiency level; growth in the forest was increased in two of the four experiments. The extra K had no effect on frost damage.

Results from 20 years of wheat experiments were used to compare responses of grain-N % and grain yield to increasing amounts of fertiliser-N. Grain-N % for both winter and spring wheat increased, mostly linearly, throughout the range, whereas many of the grain yield curves reached a maximum and then declined with further additions of N. Provided curve sections exhibiting dilution effects were excluded, the linear or nearlinear relationship between grain-N % and fertiliser-N allowed linear regression models to be fitted. This showed that, over a range of 50-175 kg N ha-l, an average of 32 kg ha-1 of fertiliser-N was required for an increase of 0.1 % N in grain dry matter of winter wheat or 56 kg N ha-1 for an increase of 1 % protein. Results from 3 years of experiments conducted by the Agricultural Development and Advisory Service (ADAS) agreed closely with those obtained from Rothamsted. Using the linear relationship between grain-N % and fertiliser-N, curves were plotted of grain yield against grain-N, adjusted for comparable additions of fertiliser-N (75, 100, 125, 150 kg ha-1). The curves tended to have sections with rising yield for small applications of N, to exhibit a maximum in a central zone and descending sections for large applications. There was no well-defined 'critical level' of grain-N%, beyond which this and grain yield were inversely related.

By adopting a uniform method of presentation, the interrelationship between grain yield, %N in grain and N uptake (or protein yield) from diverse experiments and surveys can be compared readily. All three variables can be shown together by plotting yield vs %N and adding the third variable in the form of hyperbolic lines representing equal N uptake (or protein yield). This method is used in two diagrams which display 2000 values derived from 124 experiments on winter wheat and 400 values from 41 experiments on spring wheat, conducted by Rothamsted Experimental Station over 20 years. Both scatter diagrams display well-defined boundaries which have been emphasised by linking points along these boundaries to form 'envelope curves'. For winter wheat the maximum yield was 9.5 t ha-l, the largest N concentration about 3 % and the largest N uptake 170 kg N ha-l (about 970 kg protein ha-1). Spring wheat had a maximum grain yield of 6.3 t ha-1, the largest N concentration was 3.1 % and the largest N uptake 125 kg N ha-1. Four examples of other work set within the same framework show applications of our method of presentation. Results from a recent more homogeneous series of experiments in southern England (conducted by the Agricultural Development and Advisory Service), showing clear seasonal differences, are seen to lie within the envelope curve defined by the Rothamsted experiments. Further illustrations are taken from an experiment by the Plant Breeding Institute, Cambridge, comparing varieties of different bread-making qualities, from work of the Nebraskan plant breeding team who have produced grain with larger protein content without sacrificing yield, and from survey data for England and West Germany.

Eight winter-wheat experiments conducted by Rothamsted on clay soils in 1980 and 1981 formed part of an inter-institute collaborative programme on yield variation. The tests included no N and four amounts of fertiliser N up to 200 kg ha-1, with and without spray treatments to control pests and diseases. Large grain yields-in excess of 10 t ha-1-were often obtained, mainly where losses had been limited by sprays. In Belgium, 10 experiments during the same seasons were on a wider range of soils and had spray treatments applied to all plots. Yields exceeding 10 t ha-1 were obtained at one site. Graphs showing the inter-relationship between grain-N %, grain yield and N uptake are used to compare the recent results with those from a 20-year stud) of similar data from 124 earlier Rothamsted experiments. The envelope curve delineating the range of yield and grain-N % values from these older experiments enfolded most of the more recent data points, except those deriving from the largest yields which were associated almost entirely with grain-N % values below the minimum recommended for British breadwheat. The largest grain-N uptakes were 180-190 kg ha-1 and 170-180 kg ha-1 in the recent Rothamsted and Belgian experiments respectively, compared with 160-170 kg in the older experiments. The linear or near-linear relationship between grain-N % and amounts of fertiliser N, established in the 20-year study, was again observed in the recent Rothamsted and Belgian experiments, with about 38 kg of added N needed to increase grain-N concentrations by 0.1 %.