Good question. Here is a [SWAG][1]. Well, maybe not so *scientific*.
1. Gold (Au) is exceedingly dense, ~19 g/cm<sup>3</sup>, as is mercury (Hg), ~14 g/cm<sup>3</sup>, and lead (Pb) floats on Hg.
2. Both Au and Hg are highly reflective, but Pb quickly dulls. 
3. Au is soft as butter, and the alchemists knew how [to make it into sheets 100 nm thin][2]. (One can continue the process until the gold is a transparent greenish, useful for filters but not for gilding.) Mercury is not merely soft, but **liquid**.
4. What might have been the biggest [red herring][3] is that Hg dropped onto Au dissolves into it in seconds! At first, the Hg leaves a silvery-colored stain... but in a matter of days, the stain disappears, leaving *apparently* unadulterated gold. 

   Using [Archimedes' method][4] to check purity of Au can fairly easily show if Au is adulterated with less dense Ag, ~10 g/cm<sup>3</sup>, but the more-dense Hg could be added in small amounts, appearing to increase the weight of gold without a *perceptible8 decrease in density. 

   A mountebank would think, "*Eureka!* Drop some mercury on gold, wait a few days, and *I've made gold from mercury*"... until some wise-guy heats the gold, distilling off the Hg (and *slowly* poisoning all in the assay office).


  [1]: https://en.wikipedia.org/wiki/Scientific_wild-ass_guess
  [2]: https://en.wikipedia.org/wiki/Gold_leaf#Production
  [3]: https://en.wikipedia.org/wiki/Red_herring
  [4]: https://en.wikipedia.org/wiki/Archimedes'_principle