During the course of our experiments on the lethal effectiveness of organic smokes to bacteria in the air (Baker & Twort, 1940, p. 587), we experienced, as time went on, more and more difficulty in obtaining satisfactory results, whereas in the early days of these investigations our findings were relatively consistent. In the first place, our charcoal-incense ‘candles’, when smouldered, failed to give the results expected, and in-seeking an explanation for the failure it was thought possible that the temperature of smouldering was too high and that there was pyrolysis of the emitted smoke as it passed over the glowing point of the candle. Attempts were made to elucidate the matter by (1) smouldering the ‘candle’ inverted, and (2) heating the crushed ‘candle’ on a hot plate of regulated temperature. Both methods gave somewhat improved results, but these still remained very poor, and our next procedure was to examine the effect of the smokes obtained from incense supplied from different sources. The results were irregular, as they were with several other substances which were re-tested. Only Peru balsam continued to give good ‘kills’ for some time, but, finally, we were unable to reproduce our original results even with this product. While there seemed to be evidence that the smoke, generated by smouldering on a hot plate such substances as cardboard and the incense-charcoal mix, was somewhat more lethally effective than when generated by smouldering as a ‘candle’ or strip, the experimental results as a whole had, by this time, become consistently bad.