Later attestations of 11-19 were formed by linking the digit and ten with -ir: benirmarago, linirmarago, etc. 20-90 were similar, but contracted: linago, amiago, etc. 100 was maraguin, apparently 10 with the Berber plural -en. Recco only recorded 1-16; the combining forms for 11-16, which did not have this -ir-, are included above.
Many dialects of Spanish do not distinguish [b] and [v], so been is consistent with *veen. The Berber feminine ends in -t, as in Shilha 1: yan (m), yat (f); 2: sin (m), snat (f), and this may explain discrepancies such as been and vait for 'one'.
Cairasco is a misparsed counting song, besmia mat acosomuset tamobenir marago. Ses '6' may have gotten lost in the middle of somuset (< *somussesset).
Starting with Cedeño, new roots for '2' and '9' appear ('9' perhaps the old root for '4'), new roots for '4' and '5' (arba, kansa) appear to be Arabic borrowings, and old '5', '6', '7' offset to '6', '7', '8'.