In the last quarter of the nineteenth century, three separate (but not entirely independent) papers appeared describing methods of studying complicated statistical procedures through the generation of random normal deviates. All three authors referred to problems in the smoothing of series as a motivation; all used different methods for generating the deviates. One presented itself as a method for general use and claimed to be suitable for efficient generation of large numbers of variates. The relevant works (by Erastus De Forest in 1876, by George H. Darwin in 1877, and by Francis Galton in 1890) are reproduced.
"Stochastic Simulation in the Nineteenth Century." Statist. Sci. 6 (1) 89 - 97, February, 1991. https://doi.org/10.1214/ss/1177011943