FLAMMABILITY OF PROPELLANTS 501 cidents involving aerosol containers: Explosions caused by steam, or ex- posure to sunlight, inversion of the base by direct heating, side-seam failure by solar heat. He refers briefly to other explosions in hotter climates. He knows of no fires primarily due to domestic use of aerosol dispensers. I find it hard to believe that there can have been only four such incidents, and •)ne concludes that the majority of incidents have not led to serious con- .sequences. For example, I cotfid supply the lecturer with two more from my private experience. One, an explosion of a veterinary preparation in the storage portion of a pigeon loft left closed over a weekend. This resulted in a considerable mess and frightened the pigeons but caused no fire in a second incident, an aerosol container left in a closed cabinet over a hot Bank Holiday weekend, suffered inversion of the dished base sufficiently violently to .damage adjacently stored glassware. This could have been more dangerous, (tepending on the contents of the other vessels. But even this excellent record does not supply a premise for the lecturer's conclusion. The evidence appears to show that the containers he has con- sidered are satisfactory it does not imply that leakage of the contents would present "no greater hazard than that from other common household articles". One might agree where the most hazardous component present is, say, kerosene but leakage of butane in an unventilated cupboard in which, for example, electrical switches were present, could represent a considerable hazard. It would only require the leakage of 3 oz of butane in a cupboard •)f dimensions 5'x 4'x 3' to furnish an explosive atmosphere (2% v/v). I therefore cannot entirely agree with the lecturer. MR. E. EBE•: Surely the flammability of the cretonne cloth might be influenced by other factors such as the humidity of the air? TI•E LECTURER: While I accept the value of flame temperature as a possible means of indicating hazard associated with flammable material, I feel that the duration of a flame is also of great importance and in the burst/fire tests the tendency has been for, what I suspect to be, the relatively low temperature flame produced by fairly heavy flammable material to be more likely to ignite the curtain, than the possibly higher temperature quick- flash obtained with some volatile materials. MR. E. EBEn: The flash point of the formulation is a very important property. Tests using the Abel-Pensky flash point meter with subcooled (--40øC) contents of aerosols have shown that formulations with flammable propellants have flash points of the order of below -- 40 to --30øC. But the use of fluorinated hydrocarbon propellants raises the flash point beyond room temperature. THE LECTURER: I agree that flash point is important, but its importance

$02 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS is particularly relevant to bulk storage, which is outside the scope of my paper. 3•R. E. E•EL: Why did you not compare the butane-water system with an Arcton system including some water in the solvent? From the yet un- published parts of Kempe's investigation we have evidence that a 00% volume ethyl alcohol (10% water) shows a very remarkable reduction in the flammability hazard compared with 0S).8% pure alcohol. T•E L•CTuR•: I did not compare tl•e butane/water system with an Arcton system containing some water, firstly because I feel that I have compared some products or some systems, typical of those wkich are at present being marketed or likely to be marketed in the United Kingdom, and secondly, I felt it was interesting to compare the fire suppressant properties. of water and Arcton. •. J. P. H^LL: Do the mixtures of commercial butane with water and emulsifiers form W/O or O/W emulsions on shaking? The type of emulsion can be influenced by the ratios of the two liquid phases, as well as by the emulsifying agent. Even where a W/O emulsifying agent is used, it is possible that, at low concentrations of butane, an unstable O/W emulsion may be formed. Is this so in this work, and does it affect the results? T• L•cTu•E•: The mixtures which I have considered are W/O emul- sions although O/W emulsions are known in aerosol packaging. We have been at some pains to see that those described in the paper did in fact have the water as the dispersed phase. I think instability of the emulsion could well affect the results obtained in flame projection tests, but is less likely to do so, in the case of a Burst/Fire test.