A route supported by documents
Indian decimal arithmetic reached Arabic scholarly traditions and, through Arabic-to-Latin textual work, medieval Europe. Menso Folkerts traces this history through computational writings and manuscript witnesses. The familiar label “Hindu-Arabic numerals” points to a history of adaptation as well as origin; it should not be read as the work of one inventor or one journey. 1
A crucial example is al-Khwārizmī’s arithmetic, usually dated around 825. Its Arabic original is not known to survive. André Allard’s study therefore examines twelfth-century Latin adaptations as evidence for the work and its reception. A surviving witness can be indispensable while remaining different from the lost original it helps us understand. 2
Translate a procedure, not just a numeral
Here is an original classroom calculation, not a reconstruction of a particular manuscript page. Multiply 248 by 16. In modern decimal notation:
248 × 16 = 248 × (10 + 6) = 2,480 + 1,488 = 3,968.
The procedure depends on the value of a position, not the shape of a written digit. Changing from 248 to २४८ leaves the calculation unchanged. Changing the base would change how carrying works, even if the same marks were reused.
Suppose a reader understands the digits but does not know why the partial product 248 × 10 shifts one place. A translation that gives only equivalent symbols has not yet transmitted the method. It must also convey the roles of positions, intermediate results, and carrying.
For instance, in 248 × 6, eight units times six make forty-eight units. Record eight units and carry four tens. Four tens times six, plus those four tens, make twenty-eight tens. Record eight tens and carry two hundreds. Finally two hundreds times six, plus two hundreds, make fourteen hundreds. The result is 1,488. Explaining these invariants makes a procedure portable across scripts.
What would establish a connection?
Consider two texts that contain the same numerical answer. That resemblance is a starting point, not a conclusion. We would ask whether they share an unusual sequence of operations, terminology, examples, or even a distinctive error. We would also need a plausible chronology and evidence that a text or its users could have reached the later setting.
These are methodological questions for this reference site, not a claim that every historical connection satisfies one simple checklist. A copying error may be powerful evidence in one case; in another, a common algorithm may arise independently because it is an efficient solution to a common problem.
A useful reading exercise is to make two columns: what a document explicitly records, and what we infer from it. “A Latin adaptation survives” belongs in the first column. A detailed description of the exact wording of a lost Arabic original does not follow automatically. Where a reconstruction is proposed, it needs its own argument.
More than a single outward arrow
This first edition offers selected examples, not a complete map of mathematical exchange. Language also matters within the subcontinent: compare the explanatory setting of Yuktibhāṣā with the problems in Līlāvatī. Zero and place value helps separate a written symbol, a positional practice, and arithmetic rules. Following those distinctions gives a richer history than attaching every development to one modern national border.