Short answer. F1 = parent A x parent B. F2 = F1 x F1 (siblings, in cannabis). F3 = F2 x F2, and so on. The textbook promise ("F1s are uniform") holds only when both parents are inbred. Cross two heterozygous clones, which is what most cannabis "F1s" are, and you get variation from the first generation.
What's happening
The textbook case (inbred parents)
- P1 is AA BB CC..., P2 is aa bb cc... (both homozygous).
- F1 is Aa Bb Cc... at every differing gene: genetically identical siblings, uniform, often vigorous (hybrid vigour) [2].
- F2: each gene segregates 1:2:1, so variation is maximal. With 10 differing genes there are 3^10 = 59,049 genotype combinations [1].
- F3 and later: with sibling mating and selection, heterozygosity falls and families become more uniform.
The cannabis case (heterozygous parents)
Most elite cannabis clones are highly heterozygous polyhybrids [5]. Crossing two of them is genetically like an F2 at many genes: the "F1" segregates. That's why one pack of "F1 hybrid" seed can show several distinct phenotypes. Calling it F1 is correct by definition (first generation of that cross) but misleading about uniformity.
How fast does heterozygosity fall?
| Mating system | Heterozygosity left after each generation | After 4 generations | After 6 generations |
|---|---|---|---|
| Selfing (S1, S2...) | x ½ | ~6% | ~1.6% |
| Full-sib mating | H(t) = ½H(t-1) + ¼H(t-2); tends to x 0.81 per generation | ~41% | ~27% |
| Random mating in a large population | ~no change |
Values are single-locus expectations starting from unrelated parents [2]; selection and linkage change them in practice. The point: sib-mated cannabis lines take many generations to approach true breeding, and "F5 IBL" says more about effort than about measured uniformity.
Worked example: chemotype in F1 and F2
THC-dominant inbred (B_T/B_T) x CBD-dominant inbred (B_D/B_D) [3]:
- F1: all B_T/B_D, all mixed type II. Uniform for chemotype.
- F2: 1 type I : 2 type II : 1 type III.
- F3 from a type III F2 pair: all type III. Fixed for that trait.
Note that being fixed for chemotype says nothing about the other few thousand genes.
Naming conventions
| Label | Meaning | Uniformity to expect |
|---|---|---|
| F1 (inbred parents) | Uniform hybrid | High |
| F1 (clone x clone) | First cross of two heterozygous plants | Low to moderate |
| F2 | F1 sibling cross | Lowest; best for finding new combinations |
| F3-F6 | Further sib generations with selection | Rising |
| BX1, BX2 | Backcrosses to one parent | Moving towards the recurrent parent |
| S1 | Selfed | Variable; reveals recessives |
| IBL | "Inbred line", undefined generation count | Claim; check data |
Step by step: making F2s that mean something
- Grow out at least 10-20 F1s (both sexes for regular seed).
- Select the best female and the best 1-3 males (by progeny tests if you can).
- Make the F2 cross and record parents as GB-IND IDs, not just the line name.
- Grow a large F2 (see pheno hunting for numbers) and select.
- Advance by crossing selected F2 siblings; to stabilise, keep the population from bottlenecking to a single pair every generation (inbreeding depression) [4].
Common mistakes
- Expecting uniform F1s from two clone-only parents.
- Selling F2s as if they were stable. They are the opposite.
- Making every generation from one male and one female. That is close to selfing in its inbreeding effect, without the planning.
