11. labi-n-isa = over 1.
  12. labi-n-dalaua = over 2.
  13. labi-n-tatlo = over 3.

A precisely similar method of numeral building is used by some of our Western Indian tribes.  Selecting a few of the Assiniboine numerals[98] as an illustration, we have

11. ak kai washe = more 1. 12. ak kai noom pah = more 2. 13. ak kai yam me nee = more 3. 14. ak kai to pah = more 4. 15. ak kai zap tah = more 5. 16. ak kai shak pah = more 6, etc.

A still more primitive structure is shown in the numerals of the Mboushas[99] of Equatorial Africa.  Instead of using 5-1, 5-2, 5-3, 5-4, or 2d 1, 2d 2, 2d 3, 2d 4, in forming their numerals from 6 to 9, they proceed in the following remarkable and, at first thought, inexplicable manner to form their compound numerals: 

   1. ivoco.
   2. beba.
   3. belalo.
   4. benai.
   5. betano.
   6. ivoco beba = 1-2.
   7. ivoco belalo = 1-3.
   8. ivoco benai = 1-4.
   9. ivoco betano = 1-5.
  10. dioum.

No explanation is given by Mr. du Chaillu for such an apparently incomprehensible form of expression as, for example, 1-3, for 7.  Some peculiar finger pantomime may accompany the counting, which, were it known, would enlighten us on the Mbousha’s method of arriving at so anomalous a scale.  Mere repetition in the second quinate of the words used in the first might readily be explained by supposing the use of fingers absolutely indispensable as an aid to counting, and that a certain word would have one meaning when associated with a certain finger of the left hand, and another meaning when associated with one of the fingers of the right.  Such scales are, if the following are correct, actually in existence among the islands of the Pacific.

  BALAD.[100] UEA.[100]

1.  parai.                           1.  tahi.
2.  paroo.                           2.  lua.
3.  pargen.                          3.  tolu.
4.  parbai.                          4.  fa.
5.  panim.                           5.  lima.
6.  parai.                           6.  tahi.
7.  paroo.                           7.  lua.
8.  pargen.                          8.  tolu.
9.  parbai.                          9.  fa.
10.  panim.                          10.  lima.

Such examples are, I believe, entirely unique among primitive number systems.

In numeral scales where the formative process has been of the general nature just exhibited, irregularities of various kinds are of frequent occurrence.  Hand numerals may appear, and then suddenly disappear, just where we should look for them with the greatest degree of certainty.  In the Ende,[101] a dialect of the Flores Islands, 5, 6, and 7 are of hand formation, while 8 and 9 are of entirely different origin, as the scale shows.

   1. sa.
   2. zua.
   3. telu.
   4. wutu.
   5. lima
   6. lima sa = hand 1.
   7. lima zua = hand 2.
   8. rua butu = 2 x 4.
   9. trasa = 10 — 1?
  10. sabulu.