I would argue for one-based counting [a] and zero-based indexing.

Illustrated example of an array- indexes along the top (where the data can be found), and counting along the bottom (how many values have been read):

  0   1   2   3
  | A | B | C | D |
      ^   ^   ^   ^
      "1" "2" "3" "4"

[a] https://en.wikipedia.org/wiki/Ordinal_number_%28linguistics%...