It may now be regarded as proved by Bubnov that our present numerals are
derived from Greek sources through the so-called Boethian “apices,”
which are first found in late tenth century manuscripts. That they were
not derived directly from the Arabic seems certain from the different
shapes of some of the numerals, especially the 0, which stands for 5 in
Arabic. Another Greek form existed, which was introduced into Europe by
John of Basingstoke in the thirteenth century, and is figured by Matthew
Paris (V. 285); but this form had no success. The date of the
introduction of the zero has been hotly debated, but it seems obvious
that the twelfth century Latin translators from the Arabic were
perfectly well acquainted with the system they met in their Arabic text,
while the earliest astronomical tables of the thirteenth century I have
seen use numbers of European and not Arabic origin. The fact that Latin
writers had a convenient way of writing hundreds and thousands without
any cyphers probably delayed the general use of the Arabic notation.
Dr. Hill has published a very complete survey of the various forms
of numerals in Europe. They began to be common at the middle of the
thirteenth century and a very interesting set of family notes concerning
births in a British Museum manuscript, Harl. 4350 shows their extension.
The first is dated Mij^c. lviii., the second Mij^c. lxi., the third
Mij^c. 63, the fourth 1264, and the fifth 1266. Another example is given
in a set of astronomical tables for 1269 in a manuscript of Roger
Bacon’s works, where the scribe began to write MCC6. and crossed out
the figures, substituting the “Arabic” form.