Source: an anonymous manuscript preserved in the Padua conservatory, thought to be mid-18th century. (Reproduced here below)
“To tune the harpsichord
C-G; G-D; D-A; A-E: Fifths. It is appropriate that they are narrower than the exact fifth, which means that the higher string has not to be pulled up to the right consonance but instead is left quite loose and flat.
C-F; F-Bb: Wider fourths
Bb-Eb; Eb-Ab: Narrower fourths
E-B; B-F#; F#-C#: Perfect fifths
C#-G#: Test. so it makes a perfect fifth
In this partition, major thirds are a little altered in the plus and minor thirds in the minus”
[Translation: Fabrizio Acanfora]
As with almost all mollified meantones, the instructions are somewhat open to interpretation, but the logical limits leave hardly any room to move. If the first fifths are not significantly tempered, there is no requirement for the two larger-than-pure fifths. One extreme is to make the first major third C-E pure (upper graph). The other extreme can be constructed by pushing the amount of tempering for the wide fifths to the minimum which can be heard, somewhere between 1 and 2 cents. This would allow the narrow fifths to become tempered by 1/5th syntonic comma, as shown in the lower graph. Any less than this is simply not credible. The interesting thing about this last option is that it begins to approach a sort of Vallotti on Bb. This may well be indicative of a certain general trend in the Veneto during the first half of the 18th century.
Disclaimer: the attribution of comma fractions to the tempered fifths is done by an automatic table look-up function. If the exact value is not in the table, the system grabs the nearest lower value. Naturally, my table does not have all possible fractions of both commas for both narrow and wide fifths. Therefore, with all modified meantone temperaments, for those fifths which “close the circle”, i.e. fill in the gap left by the departure from pure meantone logic, the amount of tempering has been somewhat arbitrarily chosen, exactly as one does when tuning an instrument, although in this case, working with decimal values for the proportion of the fifth. Therefore, the final size may not represent any fraction of any given comma, and there is no guarantee that the amount of tempering indicated as a comma fraction in the graphic is precise. The value indicated in cents, however, is derived from the proportion, and is therefore completely trustworthy.