For a study of the process of formaldehyde disinfection, cotton threads were used as test objects. These were, in most experiments, impregnated with a heavy suspension of a coagulase-negative micrococcus in 1% gelatin solution or in 90% horse serum, and dried. In other experiments Mycobacterium tuberculosis, Bacillus subtilis or smallpox virus was used. After exposure to formaldehyde, quarter-inch lengths of the threads were cut off and cultivated on agar plates; individual threads were recorded as fertile or sterile and the mean number of organisms per thread was estimated statistically. Neutralization of the formaldehyde was found to be unnecessary except when strong concentrations were used.

In laboratory tests there was a linear relation between the concentration of formaldehyde and the killing rate. Little effect was observed on the speed of disinfection from variation in temperature over the range 0–30°C. There was some suggestion of an optimum relative humidity for disinfection at about 80–90%, but no great increase in disinfection rate was obtained by increasing the relative humidity above 58%. Threads impregnated from a suspension of the cocci in 90% serum were disinfected much more slowly than those impregnated with cocci suspended in a solution of gelatin. Covering the threads with several layers of blanket also slowed the killing rate considerably.

Tests were made of the sterilization of the threads in full-size disinfectors. Mattresses and other bedding absorbed a large quantity of formaldehyde so that the amount supplied had to be greatly in excess of that estimated from laboratory tests to provide a lethal concentration. A disinfector working at ambient temperature was found to kill over 99.99% of the micrococci on threads exposed on the surface of mattresses. The destruction of micrococci on threads placed under pillows or blankets was less, as was that of tubercle bacilli on fully exposed threads.

A disinfector in which the air temperature could be raised to about 100°C. was found to be more effective, partly because the increased temperature allowed much greater penetration of fabrics by the formaldehyde vapour; but the relative contribution of the heat and the formaldehyde to the disinfection process could not be fully determined.

In both the laboratory and the field experiments it was found difficult to obtain completely reproducible results. This constitutes one of the most disconcerting features of disinfection by formaldehyde. The process is influenced by so many different factors that it is impossible to predict with real assurance the result under any given set of conditions.

It is concluded that disinfection by formaldehyde vapour should be used only when no other method is available. It certainly cannot be recommended for disinfection of fabric contaminated with smallpox virus or with anthrax spores, when complete destruction has to be ensured; and unless carried out with special care it is not really suitable for woollen garments and toys soiled with tubercle bacilli.