Since the last report to sponsors circulated in mid-1967, considerable concern has been felt over the wide scatter of results. The scatter in results between specimens produced in a single batch was generally comparatively small, in the order of 15%, but occasional results occurred which ehow6d much greater differences from the average. Identical experiments carried out in different batches exhibited rather more scatter than with in single batches. Similar scatter was noted in the results of other workers" 2 . Additionally, many blasting variables which had previously been considered important by metal sprayers appeared to have very little effect on the bond strength of flame sprayed aluminium coatings on mild steel. Consideration of scatter in results suggested four possible reasons for this scatter.