A revised attempt at reconstruction
This is an attempt to reconstruct the tengwar numeric mode seen in the King's Letter III (DTS 49). It contains tengwar for the numbers 1, 3, 4, and 6. They look very different from the usual tengwar numerals transcribed by Christopher Tolkien in 1981-1982 and reproduced in Quettar No. 13 (1982), reprinted in Beyond Bree Dec. 1984, an in Quettar Special Publication No. 1 1987). It may be hypothesized that they represent a particular "Mode of Gondor". At least the order (or "endianness") seen in this specimen is consistent with the transcriptions of CJRT; the least significant digit is written first.
The numerals for 6, 3, 4 and 1 look roughly as follows:
The line below is filled in with my conjectured values for the other numerals, 1 - 9. The values for 2 and 5 are "interpolated" from 1 and 3, and 4 and 6, respectively. The value for 7 is in line with the values for 1 and 4, gleaning at the values for 1, 4 and 7 in CJRTs transcriptions. The values for 8 and 9 are similar to the values in CJRTs transcriptions, slightly modified to fit the pattern of 4,5 6.
Note: All these tengwar are found elsewhere. For example, 5 and 6 are also found as Valmaric letters in PE 14 p. 99, the two rightmost columns.
One particular problem remains unsolved here; the numeral for zero. Since the current canonical tengwa for zero resembles the tengwa for 1 in this specimen, no attempt to guess what a tengwa for zero should look like in this mode.