Short answer. A breeder sets a goal, grows big populations, selects both females and males, tests what the parents actually pass on, stabilises the line over several generations, test-grows every batch and publishes its herm rate. A pollen chucker crosses two popular cuts (or selfs one), gives the seed a new name and sells it without growing it out. The first is plant breeding as every textbook describes it [1][2][3]. The second is seed production. It isn't a matter of taste: the genetics explain why one gives you a line and the other gives you a lottery ticket.
House position: we're snobby, and proud of it. GenetixBay holds breeders to the standards of plant breeding, not of seed marketing. Males are a necessity, and they're encouraged and rewarded here. Selfing is mostly frowned upon. Every line publishes a herm rate from tester grows, and lines above 2 % in tester grows don't qualify for Bay Lines, the Seed Club or featuring. We criticise practices, never named people: this article names no one, and neither should you when you quote it.
What real breeding is
Plant-breeding texts agree on the core loop: define objectives, build or choose variable populations, select, evaluate the selections' offspring, and repeat until the line breeds true for what you want [1][2][3]. In cannabis terms:
- A defined goal. "A terpinolene-dominant, mould-tolerant, 9-week line that stays female under stress" is a goal. "Cross the two cuts everyone wants" is a marketing plan.
- Large populations. Rare combinations only turn up if you grow enough plants. If a desirable type is 1 in 16 in an F2 (two recessive genes), you need about 47 plants for a 95 % chance of seeing even one; at 1 in 64 (three genes) you need about 191 [5]. A pack of ten seeds is a peek, not a hunt. See selection and pheno hunting.
- Selection of females and males. Half of every seed's genes come from its father. You can't see a male's resin or flavour directly, so breeders judge males by vigour, structure, stem-rub aroma, sex stability, and above all by what their daughters turn out like.
- Progeny testing. A plant's own looks are only part of the story; what counts is what it passes on. Breeders grow out the offspring of candidate parents and keep the parents whose families are good and uniform [1][3][4].
- Stabilisation over generations. F2s and later generations, backcrosses and inbred lines (IBLs) fix the wanted traits so the seed breeds true. That takes years. See backcrossing, IBLs and stabilisation and F1, F2, F3 explained.
- Test grows. Every batch is grown out, by the breeder and by independent testers, before release. Germination, uniformity, flowering time and sex stability are measured, not assumed.
- Honest herm-rate disclosure. The batch's herm rate goes on the record, with the number of plants it came from. Intersex plants are culled and never used as parents. See diagnosing hermaphrodites.
- Records and naming ethics. Every plant, cross and batch is traceable to its parents; the name says what it is (F1, F2, BX, S1, feminised); a famous name is never put on something it isn't. On GenetixBay that means GB-IDs with evidence levels, and documented males in the Stud Book, the male showcase where each male's selection journal, keeper and sire credits are kept. See record-keeping and naming ethics and documenting lineage evidence.
Breeder vs pollen chucker
| Breeder | Pollen chucker | |
|---|---|---|
| Goal | Written down before the first cross | "Cross the two cuts that are trending" |
| Population size | Dozens to hundreds per generation | Whatever fits in the tent |
| Males | Selected, tested, kept; the father matters as much as the mother | Whatever male was around, or none (reversed female) |
| Selfing | A last resort to save a line, labelled S1 | The default, because it needs no male |
| Testing parents | Progeny tests: judged by what their offspring do | Judged by the parent cuts' reputation |
| Generations before release | Several (F2+, backcrosses, IBL work) | One. The F1 (or S1) is the product |
| Test grows | Every batch, in-house and independent testers | None, or "the customers are the testers" |
| Herm rate | Measured, published with the plant count, culled from the line | Not measured; herms blamed on the grower's light leak |
| Name | Unique, with generation and type on the label | A mash-up of the parents' famous names |
| Records | Traceable pedigree and batch records | "Trust me" |
| What you get | A line that grows the way the description says | A segregating lottery ticket |
Why the difference shows in your tent
Crossing two clones doesn't make a uniform F1
The uniform F1 hybrids of maize or tomatoes come from crossing two inbred (highly homozygous) parents: every seed gets the same pair of gametes, so every plant is the same [1][3]. Elite cannabis clones are not inbred; they're heterozygous selections, often from polyhybrid backgrounds. Each clone makes many different gametes, so a cross between two of them already segregates like an F2 [5]. That's why a pack of "Famous Cut × Other Famous Cut" throws a dozen different plants, and why the "F1" on the pack says nothing about uniformity. Making true F1 hybrids in cannabis needs inbred parent lines first, which is exactly the slow work Barcaccia and colleagues set out as a breeding scheme [6].
Selecting only females halves your progress
The breeder's equation predicts the response to selection:
R = h² × S
R is the gain in the next generation, h² the narrow-sense heritability of the trait and S the selection differential (how far your chosen parents sit above the population average) [4][5]. When the two sexes are selected differently, S is the average of the two: S = (S♀ + S♂) / 2 [4].
Worked example: a trait with h² = 0.3. You pick females 4 units above average. - Male also selected, 4 units above: S = 4, R = 1.2 units per generation. - Random male (S♂ = 0): S = 2, R = 0.6 units. Half the progress, every generation.
That's the arithmetic behind "males are a necessity". A breeder who never selects a male is throwing away half the gain, and taking pot luck on half the genome.
Selfing and reversal concentrate what's there, good and bad
Selfing halves heterozygosity each generation (F rises from 0 to 0.5, 0.75, 0.875...), so recessive faults, including inbreeding depression, surface fast [13]. That can be a tool when a breeder does it deliberately and culls hard. Done casually, on a plant chosen because it's popular, it fixes whatever that plant carries. See S1s and selfing.
Intersex expression is heritable, and it responds to selection
This is where pollen chucking does the most damage.
- Sex expression has a genetic basis. In monoecious hemp, the degree of male and female flowering varies between cultivars and has a genetic basis [10]; genome-wide work has mapped loci associated with sex expression [11]. (Hemp, not drug-type cannabis: we're extrapolating, carefully.)
- It responds to selection, in both directions. Monoecious hemp exists because 20th-century breeders selected for plants with both sexes [8][10]. If selection can push a dioecious species towards monoecy, careless selection can drift a line towards herming, and careful selection can hold it female.
- Herm pollen makes feminised seed. Seed from hermaphroditic pollen in drug-type cannabis gave only female progeny [9]. A plant that herms in the room and seeds itself has made a new batch of "feminised" seed carrying the same tendency. Reversing an unproven female with silver does the same thing on purpose [12].
Put together: selling seed from a line no one has test-grown, or from a plant that hermed, spreads intersex tendency. The house position is that herm-prone lines are mainly a breeding failure; see hermaphrodites.
Small test grows can't prove sex stability
How many plants does a test grow need? If a line's true herm rate is p, the chance of seeing zero herms in n plants is (1 − p)ⁿ.
| True herm rate | Chance of zero herms in 10 plants | Plants for a 95 % chance of catching at least one |
|---|---|---|
| 20 % | 11 % | 14 |
| 10 % | 35 % | 29 |
| 5 % | 60 % | 59 |
| 2 % | 82 % | 149 |
And the reverse: if you see zero herms in n plants, the 95 % upper bound on the true rate is about 3/n (the "rule of three"). Zero in 10 plants only tells you the rate is probably below about 26–30 %. To back a claim of "under 2 %" with a clean test, you need about 150 plants across testers. That's why GenetixBay asks for the plant count next to every herm rate, and why tester grows are pooled.
How to spot a pollen-chucked line
- The pedigree is two famous cut names and nothing else: no generation, no male, no breeder notes.
- The first release is an F1 or S1 of popular clones, with no test-grow reports.
- No herm rate, or a herm rate with no plant count.
- Huge numbers on the pack (yield, THC) and nothing about uniformity or flowering-time range (see buyer beware).
- A new name every season for similar parents.
- When herms are reported, the answer is always "light leak".
None of this proves anything about a particular seller. It tells you what evidence to ask for.
How to do it properly (the short version)
- Write the goal down.
- Grow enough plants to find what you're after; select males as carefully as females, and keep them (register them in the Stud Book).
- Progeny-test your candidate parents before committing to them.
- Stabilise across generations; label each one honestly.
- Test-grow every batch with independent testers; cull intersex plants; publish the herm rate with the plant count.
- Keep the records; name it honestly; let the line earn its reputation.
Common mistakes
- Treating the first cross as the finished product.
- Using whichever male is available, or skipping males entirely.
- Selecting on one plant's looks without testing what it passes on.
- Quoting "no herms" from a handful of plants.
- Keeping a herm because it's the best smoke in the room.
Evidence
- The breeding loop (objectives, variable populations, selection, progeny testing, stabilisation, testing): Established plant-breeding practice [1][2][3].
- Breeder's equation and the halving of response when one sex is unselected: Established quantitative genetics [4][5].
- Crosses of heterozygous parents segregate; uniform F1 hybrids need inbred parents: Established [1][3][5]; proposed for cannabis [6].
- Selfing raises homozygosity and exposes recessive load: Established [13].
- Sex expression has a genetic basis and responds to selection: Supported in hemp [10][11], and by the deliberate breeding of monoecious hemp [8]; a heritability estimate for intersex tendency in drug-type cannabis hasn't been publishedbeing reviewed.
- Herm pollen gives all-female, herm-carrying seed: Supported in drug-type cannabis [9][12].
- Strain names hiding genetically different plants: Supported [14].
- Herm-prone lines being mainly a product of poor breeding practice: GenetixBay house position, consistent with the above and with long breeder experience; drug-type cannabis has been dioecious through recorded history [7][8].
- Test-grow sample sizes: Established binomial arithmetic.
