About BarkMath

BarkMath publishes free calculators and guides for dog owners. What makes it different from the rest of the category is not the arithmetic. It is that the model behind the arithmetic is published, tested and openly limited, and that every factual claim on the site has been checked against a primary source before it was written.

Why this site exists

Search for almost any question about your dog and you will get a number. What you will almost never get is where the number came from.

Take the most common one: a puppy doubles its weight at 16 weeks. That rule is roughly right for a medium-sized dog and meaningfully wrong for everything else, because the age at which a dog reaches half its adult weight moves later as adult size increases, about 15 weeks for a toy breed, about 21 for a giant. Dozens of calculators apply the one rule to every dog anyway, and none of them say so.

The same pattern runs through pet health content generally. Figures get repeated until they sound authoritative, and tracing them back often finds nothing underneath. Three examples we checked and could not stand behind:

• "Senior dogs need 20% fewer calories." The figure traces to a 1998 review that describes it as an assumption. A later meta-analysis of 713 dogs found no independent effect of age on maintenance energy requirement.
• "Rapid growth causes osteochondrosis." Contradicted by the standard pathology review of the condition. Overnutrition and joint disease are genuinely linked; that specific causal claim is not the mechanism.
• "Exercise a puppy five minutes per month of age." Widely published, with no traceable source of any kind. We say so rather than repeat it.

A number you cannot check is not much use. BarkMath is built on the opposite premise: show the method, show the sources, and say plainly where the answer should not be trusted.

How the numbers are made

The puppy size calculator runs a Gompertz growth curve whose two parameters (the age at which a dog reaches half its adult weight, and the growth rate constant) both vary continuously with how large the dog will eventually be. That design choice is the substance of the tool. The usual alternative sorts dogs into five size bins, which makes a 51 lb dog and a 99 lb dog share one curve while a 49 lb dog gets a different one, and produces a visible jump in the estimate at round numbers. Growth does not work that way.

The curve was fitted by least squares against percentage-of-adult-weight tables assembled from published breed charts spanning Chihuahua to Great Dane. Tested end to end, it recovers each chart's own stated adult weight with a mean absolute error of 6.9% and a bias of −1.3%. For context on what that is worth: different published charts for the same breed disagree with each other by 10–30%, so this is close to the ceiling the source data allows.

Agreeing with the charts you were fitted against is a weak test, so the model was also checked against a source it was never trained on. FEDIAF's 2025 nutritional guidelines publish independent growth curves as log-linear equations by mature-weight band. For a 30 kg dog at 20 weeks, FEDIAF gives 54.5% of mature bodyweight; our model gives 51.6%, under three percentage points apart, from data with no shared lineage.

Breed figures come from measured population data where it exists: 74 of the site's 268 breeds and varieties carry sex-specific adult weights taken from a study of over 114,000 young adult dogs. Those are real population means, not breed-standard show ideals, which matters because pet dogs commonly sit outside the standard.

All of it (the equations, the residuals by size band, the calibration table and the cases where the model performs worst) is on the methodology page. It is there so you can disagree with it.

The editorial standard

Across 16 published guides, the same four rules apply without exception.

• Every factual claim is sourced, and linked where the claim is made. Not a bibliography at the bottom that nobody maps back to the sentence it supports: the link sits at the point of the claim, and the full source list appears at the end as well.
• Verify before citing. A citation only goes in if the page behind it has actually been opened and read end to end. If a fact cannot be verified against a primary source, it is either left out or published with the uncertainty stated. "We could not verify this" is a legitimate and frequently useful thing to publish.
• Limitations are stated, not smoothed. The most rigorous published growth standards exclude dogs over 88 lb, so our giant-breed estimates carry a wider range and an explicit warning. Our medium band fits about five points low, because the available charts for that weight lean on early-maturing brachycephalic breeds. Both are disclosed in the interface, not buried.
• For anything emergency-adjacent, the standard goes up. On toxicity or dosing content, phone numbers come before the page's own numbers, and "call before you act" comes before anything a worried owner might try to self-manage with.

Corrections are welcome and are acted on. If something here is wrong, or a source does not say what we claim it says, tell us. That is the most useful email this site receives.

Who writes this

BarkMath is written, built and maintained by Michael Evans, based in London, Ontario, Canada.

I am not a veterinarian, and I hold no veterinary or animal science qualification. It is worth being direct about what that does and does not mean.

It means BarkMath does not offer clinical judgment, and nothing here is a substitute for an examination by someone qualified to give it. There is no veterinarian on staff and no veterinary review step, and the site does not claim one.

What the site does offer is the other half of the work: finding the primary literature, reading it properly, representing it accurately, building a model that can be tested against it, and publishing the method and the failure modes so a reader, or a vet, can check it. The claim being made is not authority. It is traceability.

Guides carry a "BarkMath Team" byline. That is a brand decision, and it should not be read as implying a newsroom. The site is one person, named above, and the accountability for anything published here is mine.

How BarkMath is paid for

BarkMath is free to use and currently carries no advertising. It is intended to be supported by display advertising, which is a deliberate choice over the alternatives: advertisers have no involvement in and no advance sight of what gets published.

There are no affiliate links anywhere on this site, no sponsored posts, and no paid product placements. No brand has ever paid for a mention, and no article recommends a specific product in exchange for anything. If that changes, it will be disclosed on this page and on the article in question before it goes live, not afterwards.

What BarkMath is not

• Not veterinary advice. It cannot examine your dog, and no statistical estimate can.
• Not a dosing tool. Never calculate medication, anaesthetic or feeding amounts from a predicted weight. Use your dog's actual measured weight for that, confirmed by your vet.
• Not an emergency service. If your dog is unwell, has eaten something it should not have, or is growing very differently from what you expected, phone your veterinarian or an emergency line. Do not wait on an email.

The site at a glance

Operated by	Michael Evans, London, Ontario, Canada
Contact	hello@barkmath.com
Breeds and varieties covered	268
Published guides	16
Model accuracy	6.9% mean absolute error, −1.3% bias
Method	Published in full
Veterinary review	None. See above.