Until the invention of the tuning fork in the late 18th century, there was no absolute way to specify a reference pitch, since there was no way to either measure or produce a specific frequency. Whenever an external reference is mentioned at all, old tuning methods all but universally instruct the reader to take the starting note from a wind instrument, since only wind instruments were limited in their ability to adapt to slightly higher or lower pitch levels. Strings could tune higher or lower by at least a quarter-tone without either breaking strings or greatly altering the overall behavior of the instrument.
Why piano tuners never have any fun!
Naturally, if you are going to take a note from a wind instrument, you want a note which meets several criteria:
1. If the point is to make sure the wind instrument can play well together with the keyboard in all keys, it should be one of the least flexible notes on the instrument, i.e. a note which does not allow for significant amounts of pitch change, either by variations in pressure/embouchure or by alternative fingerings
2. It should be a note which is found in a majority of the most often-used keys
3. It should be a note which provides a practical starting point according to the logic of the temperament
As the chart below clearly shows, the historically reference note has overwhelmingly been C, with F as a distant second. These notes are the “home key” tonics of the bassoon and oboe, the most often-used wind instruments in mixed ensemble playing. Bassoon is especially important, since its task is to double the bass line as played by the keyboard (or lute). Therefore, the bassoon was required to precisely mimic whatever the temperament was, whereas the oboe – like all melody instruments and the voice – was expected to more or less freely deviate from the keyboard temperament in order to favor pure tuning whenever possible.
The switch to A seems to have come around the middle of the 19th century. This coincides exactly with the wide-spread adoption of Equal Temperament, as well as the introduction of truly chromatic wind instruments, i.e. instruments with complex key systems which removed both the instability and the contrast of color which characterized the accidentals on earlier winds. Naturally, in a context of a perfectly-even distribution of the 12 tones of the octave, no one reference note is any better than any other. Furthermore, since variation in temperament was eliminated, string players no longer needed to take the tuning of each and every one of their strings directly from the keyboard in order to match the degree of tempering among the fifths. Instead, they could simply take a single note according to which the rest of their strings were tuned as either ever-so-slightly narrow fifths or simply as pure fifths, in which case the maximum error would be a mere 4 cents.
An instructive illustration of the change in mentality is found in Carl Kützing’s 1833 treatise on piano construction. His appendix on tuning the instrument begins with a discussion of temperament in general and the primary causes for the need for temperament, explaining both the Pythagorean and the Syntonic comma. These explanations are accompanied by various tables of proportions and hypothetical series of intervals, all of which follow the traditional approach of reckoning everything from a starting point of C. However, when it comes to the actual tuning instructions, the first step is to tune bass A from tenor a as a pure octave, implying that tenor a was the reference note which had been established either by using a fork or a wind instrument. The rest of the instruction proceeds according to the usual 18th century manner of producing Equal Temperament, i.e. the major thirds A-c#-f-a are all tempered identically, and then each series of four fifths which lie between these notes are in turn tempered identically (a vestigial practice from setting meantone). It goes without saying that the same approach works equally-well (bad pun!) starting from C, or any note for that matter, so the specific use of tenor a indicated that a definite shift in methodology was already underway.