This paper seeks to explain the radical decrease in the use of the passive voice in Present-day English scientific discourse. A number of different linguistic factors having been discounted in previous research, it is hypothesised here that passives are being omitted for two reasons. Firstly, they became conventionalised in scientific discourse and subsequently lost the pragmatic function which originally justified their high frequency in scientific texts. Secondly, over the course of the twentieth century two sociocultural circumstances converge that exert pressure on conventionalised passives to disappear, namely (i) the increasing competitiveness in the scientific community, and (ii) the democratisation of discourse. This hypothesis is tested in the present paper by analysing the function of passives in scientific discourse before the drop in frequency began, that is, in Late Modern English (1700–1900). With data from ARCHER and other sources the paper will try to show that passives in Late Modern scientific English exemplify the conventionalisation and loss of contextual function of pragmatic strategies, a scenario that, given the right sociohistorical conditions, leads to linguistic change.