20 karine chemla
collective work involved in the making of a proof eventually produced and
written down by an individual, W. Th urston makes us fully aware of the
bias that such an approach represents. In fact, there are further diffi culties
linked to the nature of the sources with which the historian works.
Some of these sources, like Babylonian tablets, were discovered in
archaeological excavations, on a spot where they were used by actors.
Others came down to us through the written tradition. In most cases, the
physical medium has travelled. 34 In the end of the best-case scenario, those
that can be read are available to us through critical editions. Th rough the
various processes of transmission and reshaping of the primary sources, the
agendas related to proof described earlier may have left an imprint. In such
cases, our analysis of the source material would be biased at its root.
We shall illustrate this problem with a fundamental example, which
will bring us back to nineteenth-century historiography of proof and a
dimension of its formation that we have not yet contemplated. Above, we
outlined the contribution that this book makes to analysing the evolution
of European historiography of science with respect to ‘non-Western’ proofs.
As a complementary account, the fi rst section of Part I in the book focuses
on the approach to Greek geometrical texts that developed in the late nine-
teenth century and the beginning of the twentieth century. Th ree chapters
examine how the critical editions of Euclid’s Elements and Archimedes’
writings produced by the philologist Johann Heiberg, on which we still
depend for our access to these texts, refl ect, and hence convey, his own
vision of the mathematics of ancient Greece. Th ese chapters illustrate a
new element involved in the historiographic turn described above: the pro-
duction of critical editions. Let us sketch why they invite us to maintain a
critical distance from the way sources have come down to us, lest we uncon-
sciously absorb the agendas that shaped these editions.
Th e problem aff ecting these critical editions was fi rst exposed by Wilbur
Knorr, in an article published in 1996, the title of which was quite explicit:
‘Th e wrong text of Euclid: on Heiberg’s text and its alternatives’.^35 I n i t ,
Knorr explained why in his view, Heiberg shaped Euclid’s text on the basis
of his own assumptions regarding the practice of axiomatic–deductive
systems in ancient Greece. Knorr’s article began with a critical examination
of a debate which at the end of the nineteenth century opposed Heiberg to34 Th e research programme entitled ‘Looking at it from Asia and Africa: a critical analysis of the
processes that shaped the sources of history of science’, led by Florence Bretelle-Establet and
to which A. Bréard, C. Jami, A. Keller, C. Proust and myself contributed helped me clarify my
views on these questions.
35 Knorr 1996. A paper that appeared posthumously took up this issue once again: Knorr 2001.