Census Puzzler

[Dean Hickerson]

Hans Havermann wrote:

> 90519 sequences grouped by number of database terms:
>
> http://odo.ca/~haha/j/num/census.jpg
>
> The prominent spikes at the right of the graph are for 81 (1403 
> sequences), 90 (695 sequences), 99 (2724 sequences), and 105 (1575 
> sequences). Why these four numbers?

At one time, Neil's software restricted the %S, %T, and %U lines to 81
characters, including the "%S", the A-number, and the comma at the end.
(I think the line limit got increased slightly a while back.)  If a
sequence consists only of 1-digit numbers, as many of them do, then that
gives 35 numbers per line, which explains the 105.  I'm not sure about the
others, but they're all multiples of 3, so I'd guess that they were created
by programs with different line length limits.

Dean Hickerson
dean at math.ucdavis.edu