The starting point for this account is 1900 when J. S. E. Townsend conceived the diffusion battery technique and demonstrated it by measuring the diffusion coefficient of ions in various gases. Application
to aerosols, however, awaited developments in the fundamental physics of airborne particles. A key step was Einstein's derivation in 1905 of the equation relating the random displacement of airborne (or
liquid-borne) particles to their mobility. Cunningham and Millikan found an accurate expression relating mobility to particle size. By 1914, R. A. Millikan and his colleague, H. Fletcher, had tested Einstein's
theory by experiment with gas-borne particles and found that it was correct. Thus, aerosol particles could be expected to obey the same diffusion equations as gas-borne ions. In 1935 Nolan and Guerini described
a parallel-plate diffusion battery and used it to measure the size of Aitkin particles. Two years later, Radushkevich used a cylindrical tube battery to measure the size of a laboratory aerosol. The Nolan-Guerini
diffusion battery design was used for some 30 years without much change. The term ''diffusion battery'' appears to have been coined in 1943 and came into general use by 1960 or so. The first successful
experimental verification of the diffusion battery penetration equation was accomplished by Megaw and Wiffen in 1963. Great advances in diffusion battery design ("collimated holes structures," wire screens)
were wrought by Sinclair and his colleagues in the early 1970s. Another great advance was made in 1980 when Cheng and Yeh found that fan-model filter theory works very well for calculating penetration through
wire screens. Another milestone in diffusion battery history was cemented in 1957 when Pollak and Metnieks discovered that diffusion batteries could yield size distributions, not just the mean particle
size. This subject, called "data inversion" among other things, has been addressed in the years since by a number of mathematicians, who have proposed a number of ingenious techniques for data inversion.
The status seems to be that, given good conditions, the peaks of a bimodal aerosol can be resolved if the modal diameters differ by a factor of 2 or more.