Alright so I was never good at math but I think you have to take the indefinite integral of the creature in order to get the area of the legs under the creature, this of course does not take into account if the legs are not under the creature.
From this formalism, I presume number of legs is being evaluated based on a 2D projection of the creature in question. In this case, area under the legs is going to scale as a function of that projected area, which can me misleading. When 60ft women attack, for example, they do not do so with 100 times the legs of a 6ft women, no matter the hopes of dreams of those attacked. As such, we need to calculate the modality of the projection.
That said, we need to be careful. Sometimes, a foot is off the ground at the moment of projection, so it would be inappropriate to count the number of subdivisions in that projected area. I propose Nason & Sibson's 1992 method, which has the advantage of providing a fractional metric for "partially multimodal" distributions. This helps to avoid semantic debates such as, "Is that even a leg?!"1 Even then, however, the method will need to be modified in some special conditions, but we should get a reasonable estimate in conditions where the angle of the legs lean toward (or are parallel with) the body's center of mass.
Since 2D projection loses information, the final step is to repeat this process from a random selection of viewpoints around the creature. Provided this sampling distribution is collected with sufficient granularity, we should be able to make an educated guess about the maximum number of legs a creature may have.
-
Note: For the purposes of this analysis, load-bearing tails are legs. Non-load-bearing tails are partial legs. I will not be taking questions at this time.