Magic body size

Crossing two of them

Only heterozygotes are born wild

The founder table lists Big/n, Small/n and n/n and nothing else. It is written out rather than derived from Hardy-Weinberg, because random mating is exactly what this table is not:

private final FounderTable founders = FounderTable.builder()
        .weight(Big,   n, 40.0)
        .weight(Small, n, 40.0)
        .weight(n,     n, 20.0)
        .build();

So every doubled horse in the world, in either direction, is something somebody bred. That is the same rule the health loci use — nothing affected is ever wild-caught — pointed at something worth having instead of something to avoid. And since the percentage travels with the copy, breeding two good copies together is a real project with a real result.

Its gene carrot

Gene key

horsegenetics.body_size

Priority

140 (magical band)

Alleles

Big Small n

Combinations

6, all carryable — 6 outcomes, one each

Inheritance

codominant — both copies contribute and the percentages add

Default allele

n

In the wild

heterozygotes only — 40% Big/n, 40% Small/n, 20% n/n

Founder draws

1 × nextFloat()

Epigenetic draws

12 × nextFloat() per copy (one nextGaussian())

Per copy

normal, mean 10%, σ 7%, floor 1%

Coat effect

none, ever

The combination table

Six combinations, six outcomes. That is unusual in this model — most genes collapse several combinations onto one row — and it is exactly what codominance is: every combination genuinely differs, so sharing any two rows would be a lie about the gene.

The percentages add

Every allele copy carries a percentage, and both copies contribute: the horse's size is one plus the sum of them, Big counting positive and Small negative. Two big copies of +18% and +21% make a horse nearly 40% over — far outside anything the wild population contains.

@Override
public void contribute(AllelePair pair, Genotype genotype, GeneEpigenetics epigenetics,
                       TraitBuilder out) {
    double sum = signedDelta(pair.first(),  epigenetics.copy(0))
               + signedDelta(pair.second(), epigenetics.copy(1));
    if (sum != 0.0) {
        out.multiplyScaleUnclamped(1.0 + sum);
    }
}

Signing the contribution rather than branching on the combination is what makes the table fall out of one loop: Big/Small cancels because a positive and a negative are added, not because anything checks for it.

It needs both copies' epigenetics

Every other epigenetic gene asks for "the copy that expresses", because one locus produces one result. A codominant one cannot: asking for the expressed copy would count one allele twice and the other not at all. So this gene takes an GeneEpigenetics and asks it for copy(0) and copy(1) — the trait-side twin of the accessor mane colour needed for the same reason on the coat side.

A normal distribution, centred on a tenth

One copy's percentage is a normal draw with mean 10% and standard deviation 7%. So about two thirds of carriers sit between 1.03× and 1.17× (or 0.83× and 0.97×), and the tails reach further.

The shape matters because almost every wild horse carries this gene. If a single copy were dramatic then dramatic would be the baseline and nothing would read as unusual. Instead the population has a quiet, continuous spread of sizes — you notice it across a paddock rather than on any one horse — and the striking animals are the ones somebody bred.

A copy's percentage is also floored at 1%. A Big allele that rolled negative would be a big allele making a horse smaller — not a distribution tail, a bug in the reader's head.

Past the natural clamp, on purpose

HorseTraits.MIN_SCALE / MAX_SCALE (0.45–1.75) exist so that no amount of stacking real size loci and dwarfism can produce a horse that is not a horse. A magical gene is allowed to produce exactly that, so it applies its factor after that clamp through TraitBuilder.multiplyScaleUnclamped, bounded only by MAGICAL_MIN_SCALE / MAGICAL_MAX_SCALE (0.1–10).

With a bounded normal distribution the gene now reaches about 2.04× at the absolute limit rather than the 10× of the first draft — and it takes two good copies to get past even the natural clamp, which is the point of making it codominant. The magical bounds stay as guards rather than as the design.

The two stages compose the way you would want: the natural loci settle the horse's own size first and the magic multiplies whatever that turned out to be, so a magically enormous pony is still smaller than a magically enormous draught horse.

It needed a rendering fix

Vanilla advances a horse's leg-swing phase from the ground distance it moved and nothing else. The only size compensation anywhere in it is a hard-coded "a foal's legs cycle three times as fast"; nothing consults Attributes.SCALE, because before this gene nothing changed it.

So a scaled horse walked with its feet sliding — longer legs swinging at the ordinary rate over the ordinary ground. GeneticHorseRenderer now divides walkAnimationPos by the render scale, which makes a bigger horse take proportionally longer, slower strides. See stretchGaitToSize.

Deliberately size only

It moves no other number — not speed, not health, not jump. A large horse is a spectacle and a stable problem, not a better horse, and the natural loci already own the trade between size and everything else.

Everything visible about it comes free from one attribute: Attributes.SCALE scales the model and the hitbox.