One pairing.
Eighty-one coats.
Two parent dogs, four coat genes, and the number of genetically distinct outcomes runs into the dozens before colour is considered at all. Enter any two dogs below and see every outcome the pairing can produce - including the ones that look identical and are not.
How a Punnett square
actually works.
Each parent carries two alleles at every locus and passes one of them, at random, to each puppy. A Punnett square lays the sire's two alleles down one edge and the dam's two down the other, then fills the grid with every combination the two can make. Four cells, four equally likely draws.
The grid to the right crosses two furnished carriers at RSPO2. Each parent carries one furnishings allele and one unfurnished allele. Three of four cells produce a furnished puppy. One produces an unfurnished puppy from two furnished parents.
Run one square per locus. Because these loci sit on different regions of the genome and assort independently, the outcomes multiply rather than add. Three genotypes at one locus and three at another is nine combinations, not six. Four such loci is eighty-one.
Two furnished carriers
RSPO2 · F/I × F/I| F | I | |
|---|---|---|
| F | F/F | F/I |
| I | F/I | I/I |
50% F/I furnished, carries unfurnished, looks identical to F/F
25% I/I unfurnished, from two furnished parents
Enter two parents.
See every outcome.
Four loci, four squares, one running total. Set any locus to "not tested" and the tool will tell you plainly that no honest prediction can be offered for it. That is not a limitation of the tool. It is the actual state of any pairing where the testing was not done.
Sire
Dam
genetically distinct coat outcomes across four tested loci
Coat genetics alone: 81 distinct outcomes.
This calculator applies standard Mendelian inheritance to four independently assorting coat loci. It reports what a pairing can produce and in what proportion. It does not predict which outcome an individual puppy will receive and it is not a substitute for testing the puppy. Proportions describe a whole litter, and small litters routinely deviate from them. The colour multiplier assumes three genotypes per heterozygous locus and is offered as an illustration of scale, not as a prediction for any specific pairing. Note also that KRT71 and MC5R both report as C and T while describing unrelated traits, and FGF5 reports a T as well. Always read the gene name beside the genotype.
The most valuable answer
is impossible.
A Punnett square is usually taught as a prediction tool. For a working breeding program its more important use is exclusion. It proves which outcomes a pairing cannot produce, and that proof is what lets a breeder rule out a mating before it happens rather than explain it afterward.
Prediction gives you proportions across a litter. Exclusion gives you a hard boundary. A pairing that cannot produce a double merle is safe from that risk regardless of how the individual draws fall, and no amount of bad luck changes it.
This is also why carriers matter more than appearances. A dog that looks furnished may be F/F or F/I, and those two dogs behave completely differently in a breeding program even though they photograph the same. The square is how you tell them apart on paper before the test results confirm it.
You cannot develop a breed
you cannot predict.
Stokeshire is developing a distinct type, the Golden Australian Mountain Doodle, from Bernese, Golden Retriever, Australian Shepherd, and Poodle foundation stock. A breed in development is not a marketing phrase here. It is a commitment to a specific, uncomfortable piece of arithmetic.
Every generation of a developing type starts wide. Foundation dogs are heterozygous at most loci, which is exactly what the calculator above shows as eighty-one outcomes. Consistency is not achieved by choosing prettier dogs. It is achieved by knowing every genotype in the program, tracking which combinations narrow the range without narrowing the gene pool, and accepting that this takes generations rather than seasons.
Anyone can photograph two beautiful parents. The work is knowing what those two dogs can produce, publishing it before the litter is born, and being right.
The panel behind
every square.
Every Stokeshire breeding dog is published with its full Embark results. Every planned pairing is published with its expected split. Check our arithmetic.
References
Cadieu, E., Neff, M. W., Quignon, P., Walsh, K., Chase, K., Parker, H. G., VonHoldt, B. M., Rhue, A., Boyko, A., Byers, A., Wong, A., Mosher, D. S., Elkahloun, A. G., Spady, T. C., Andre, C., Lark, K. G., Cargill, M., Bustamante, C. D., Wayne, R. K., & Ostrander, E. A. (2009). Coat variation in the domestic dog is governed by variants in three genes. Science, 326(5949), 150-153. https://doi.org/10.1126/science.1177808
Hayward, J. J., Castelhano, M. G., Oliveira, K. C., Corey, E., Balkman, C., Baxter, T. L., Casal, M. L., Center, S. A., Fang, M., Garrison, S. J., Kalla, S. E., Korniliev, P., Kotlikoff, M. I., Moise, N. S., Shannon, L. M., Simpson, K. W., Sutter, N. B., Todhunter, R. J., & Boyko, A. R. (2016). Complex disease and phenotype mapping in the domestic dog. Nature Communications, 7, Article 10460. https://doi.org/10.1038/ncomms10460
Bauer, A., Hadji Rasouliha, S., Brunner, M. T., Jagannathan, V., Bucher, I., Bannoehr, J., Varjonen, K., Bond, R., Bergvall, K., Welle, M. M., Roosje, P., & Leeb, T. (2019). A second KRT71 allele in curly coated dogs. Animal Genetics, 50(1), 97-100. https://doi.org/10.1111/age.12743
Salmela, E., Niskanen, J., Arumilli, M., Donner, J., Lohi, H., & Hytonen, M. K. (2019). A novel KRT71 variant in curly-coated dogs. Animal Genetics, 50(1), 101-104. https://doi.org/10.1111/age.12746
Online Mendelian Inheritance in Animals. (n.d.). OMIA:002750-9615: Reduced hair shedding, MC5R-related in Canis lupus familiaris. University of Sydney. https://omia.org/OMIA002750/9615/
Veterinary Genetics Laboratory. (n.d.). Curl (C1, C2). University of California, Davis. https://vgl.ucdavis.edu/test/curl