[1] The construction of a reliable thermometer confronted early investigators with a circularity that was not merely technical but epistemological. [2] To graduate an instrument one needs fixed points — reproducible temperatures against which the scale can be anchored — yet to know that a proposed fixed point is genuinely constant, one already needs a trustworthy thermometer. [3] The boiling of water, for example, was an obvious candidate for a fixed point, but eighteenth-century experimenters found that its indicated temperature varied with atmospheric pressure, with the depth of the vessel, and even with the material of the container. [4] The apparent standard thus dissolved, under scrutiny, into a family of divergent phenomena.
[5] Historians once told this story as a straightforward march toward precision, in which better instruments progressively disclosed the true fixed points that cruder ones had missed. [6] A more searching account rejects that picture as question-begging, for it presupposes the very standard whose establishment is at issue. [7] What the early thermometrists actually did, on this reading, was not to discover pre-existing fixed points but to negotiate them: they selected, among competing candidates, those that could be made to cohere with a broader network of assumptions about heat, and they stipulated conditions — a specified pressure, a particular purity of water — under which a phenomenon would count as fixed. [8] Precision, on this view, was less a revelation than an achievement, secured by decisions that the later, tidier narrative conveniently forgets. [9] To call those decisions arbitrary would be to miss their point: they were constrained, at every step, by the demand that the resulting scale prove coherent and fruitful, even if it could never simply be read off from nature.
