The results of a yield trial may be usefully indicative; they can never be decisive. Geographical differences of soil and climate and seasonal fluctuations, limit the applicability of results to the locality and the year of the experiment. No more can be achieved than to ensure that the data accumulated in the prescribed circumstances have as high as possible a degree- of statistical probability. For this, there must be adequate precautions against a variety of factors predisposing to error. These factors may be broadly classified as follows:
(a) Soil differences within the locality, i.e. major-scale differences as, for example, between the two ends of a 10-acre field.
(b) Soil differences within small portions of the test area, e.g. as between contiguous areas of 1 square ft. or even as between the areas occupied by neighbouring single plants.
(c) Differences in the seeds sown.
(d) Damage by wire-worm, etc., resulting in the killing off of some plants especially in the very early stages of growth.
(e) Irregularities in light intensity and inter-plant competition owing to irregular seed rate or depth of sowing, etc.
(f) Losses during harvesting.
(g) Irregularities in the water content of the harvested material.