A second printing of the Summa (Toscolano, 1523) as a cross-check on the 1494 reading
What this page is: a cross-check of our reading of the 1494 print against a second printing of the same book. The 1523 reprint is not an independent witness to what Pacioli wrote. The readings are machine readings made by AI under human direction and control, and are not proofread by a person.
Machine readings, not proofread by a person. Page dated 12 October 2026.
What it is
The Summa de arithmetica was printed in Venice by Paganino Paganini in November 1494. Paganini printed it again at Toscolano, on Lake Garda, in 1523. The second printing was set from a copy of the first, so it is not an independent witness to what Pacioli wrote. It is a second compositor, working about thirty years later, with a larger and clearer type, who had to read the same abbreviations and decide what they meant. That is why it is useful.
Our copy is the ETH-Bibliothek Zürich scan (Rar 9080, DOI 10.3931/e-rara-9150, Public Domain Mark), with page images and machine-read text. Tractatus XI (the bookkeeping treatise) is on scan pages 418 to 442, folios 198v to 210v.
How we used it
It was used after the 1494 reading was made and corrected, as a cross-check. It did not supply the reading. For the whole treatise we compared our text (28,696 words) word by word with the 1523 text (28,470 words of machine reading). 17,306 words matched outright. For the words that differed, we asked whether our abbreviation signs, expanded in the usual way, give the 1523 spelling.
Result
Words
Not abbreviated in our text (spelling variants only)
2,022
Our signs expand exactly to the 1523 spelling
1,311
Fit apart from vowels or spelling
512
Not explained
411
We then looked at the page images for the not-explained and for most of the vowel-variant cases: 691 words in all. The 512 included about 240 places where our sign is the "etc." sign. These were not individually checked, because the tool's table simply lacked "etc.".
What we found
Our reading held up. Of the 691 words checked on the images, 10 needed correcting, about 1.4%. Most of the rest agreed outright, or differed in a way that did not change the word.
The corrections:
ſᵃ to ſa (p441 L34)
p̃ſto to p̃ſſo, "presso" (nearer) not "presto" (sooner) (p441 L35), which changed the translations
dimẽco to dimẽo, "dimeno" (p447 L11)
poch̃ to ꝑoch̃, "peroche" (p448 L41 and p456 L03)
õ to d̃, "de" (p455 L49)
ꝑ to ꝓ, "pro", in pro e danno (profit and loss) and "loro pro" (p444 L31, p451 L03 and L07)
cõptita to cõꝑtita, "compartita" (p455 L10)
A further change at p434 L51, aꝑtamẽte to aptamẽte, was reversed after a closer look at the 1494 print: the foot-curl is a ꝑ, "apertamente". It is not counted.
The 1523 also confirmed many ꝑ signs as "per", so the stroked p was not being systematically confused with ꝓ.
What it tells us about printing and abbreviation
The 1523 compositor spelled out much more. Where 1494 prints cᵒ, tp̃o, pᵃ, 1523 often prints cento, tempo, prima. Those are an independent second reading of what each abbreviation means.
He used a wider set of signs. For "con" the 1523 often uses the reversed-c sign ꝯ, where the 1494 print has cõ. In our reading of 1494 ꝯ appears 5 times against 546 cõ. Both printings had ꝑ and ꝓ.
He corrected some misprints silently. One example is trouodãdola in 1494 (an extra "od"), which became trouandola in 1523.
He copied some slips too. On the carried-forward balance (p456) the 1494 print has 28 lire in the closing line and 18 on the new page; the 1523 print has the same 28 and 18. The 1523 machine text has the same "g 5" that should be 12 in the Martino example on p451 (read from the machine text only, not the image).
Signs and spelling shift. Several signs sit in different places (the bar over p or over e in tempi) with the same meaning. Some 1523 words are spelled differently, such as Giouani for Giovanni, mercadanti for mercanti.
The hard words
dilimistri (p458 L13): the 1523 prints it as one word, "lane dilimiſtri", with no line break. This does not settle the meaning.
coculegni: the 1523 has "rubini coculegni desligadi", rubies coculegni unset. The meaning is still open.
pontidi: "diamanti gressi a tauola e põtidi". A tavola is probably table-cut, and pontidi is probably "pointed". We have not verified either.
Gitti and the 1523
Gitti's preface (1878) says he took the 1494 as his base and revised it "word by word" against the 1523, using the copies at the Biblioteca Marciana in Venice, and that the two editions are "in everything and for everything" identical. Two students of the Venice commercial school made the copy and compared the proofs with both editions. So Gitti is not independent of the 1523. Where Gitti and the 1523 agree, that is closer to one reading than two. His claim of identity is also too strong: the two printings differ in many details.
Limits
The 1523 is a reprint, not an independent witness. Agreement shows the letters read the same way, not that they are right.
The 1523 text used for the word comparison is uncorrected machine reading, so its letters are approximate. The page images were used for every correction.
The machines' readings were mostly not rechecked by a person.
About 240 "etc." signs were not checked individually.
Four places remain doubtful and unchanged: bẽ/bñ (p453 L19), attenẽre (p457 L55), cõ/cou (p452 L43), uigẽtini (p438 L38).
Two places where the 1523 reads differently and we kept ours: p440 L34 and p447 L09, where 1523 prints q̃ſto and Gitti has "quanto" and "quando". Also p435 L58, where the 1523 prints monete and we have mõte (monte).
Our copy of the 1523 scan is the ETH-Bibliothek Zürich e-rara scan. The evidence files are kept with the project files.