Where a lost cat actually is

When a cat gets out, the first thing an owner cannot answer is how far to search. One street? The whole neighbourhood? Several kilometres? There is published evidence for this question. There is also a gap in that evidence, and this article is explicit about where the gap is.

Distance: most cats are closer than people think

Huang et al. 2018 (Animals 8(1), 5) analysed 1,210 missing cats and reported, for cats that were found:

A median found-distance of about 50 m, with 75% of found cats within 500 m of the escape point.

It varies sharply with how the cat lives. Indoor-only cats: median 39 m, 75th percentile 137 m. Cats with routine outdoor access: median about 300 m, 75th percentile 1,609 m. Outdoor cats: median 183 m, 75th percentile 1,609 m, though that sample is small and uncertain.

So if an indoor-only cat escapes, the first thing to search exhaustively is a 140 m radius. Posting flyers across the district before doing that is the wrong order. This is not a promise that the cat is inside that circle. These are percentiles, not guarantees.

Place: gardens, undergrowth, and behind furniture

The same paper lists where cats were actually found: other people's gardens, just outside the cat's own home, bushes and shrubs, under decks, porches and terraces, garages, inside other people's houses, behind furniture, basements.

The common thread is concealed, tight, dark. Walking through open ground will not find them. You have to crouch and look underneath.

One fact matters more than the rest: a missing cat may be nearby and still not call out, and may not respond to its owner's voice, because it is frightened. Silence is not evidence of absence.

Time: 34% within a week, then a slower climb

Live-recovery rates in the same study: about 34% within 7 days, 50% within 30 days, 56-57% within 61 days, 61% within a year. Among cats recovered alive, the median time missing was about 6 days.

After 61 days the curve flattens, but cats have been found later than that. The probability does not become zero on any particular day.

How many come home on their own? Watch the denominator

Lord et al. 2007 found that of 73 recovered cats, 48 returned home by themselves.

That figure splits in two depending on the denominator. Against recovered cats only, it is 66%. Against all 138 cats that went missing, it is about 35%.

You will often see "66% of cats come home on their own". That is the share among cats that were recovered at all. It is not a statement about every cat that goes missing.

You cannot say males go further

Hall et al. 2016 reports, for normal home ranges, that males tend to range wider than females, that cats aged 2-7 range wider than those over 8, and that ranges are wider where housing density is lower. The effect of neutering is not clear.

But a normal home range and a found-distance after escape are different quantities. The following are all invalid:

Multiply the search radius by 1.88 because the cat is male. Multiply by 14.4 because the area is rural. Use a mean GPS home range directly as a search circle.

Home-range research belongs in the margins: estimating uncertainty at the far end, and deciding when to stop searching. It is not the search circle.

Beyond this point, the cat literature is silent

That is close to everything the cat research supports. And it is not enough to build a tool. Two things are missing.

Missing piece 1: what does "searched, found nothing" mean?

You checked a hedge. Nothing there. Does that mean the cat is not there? No. It may have been hidden and invisible.

Ecology has a formal answer. MacKenzie et al. 2002 (Ecology 83:2248-2255) models site occupancy when detection probability is less than one: given that you searched n times and found nothing, it computes the probability of true absence.

This is standard practice in wildlife survey work. We could not find it applied to lost-cat search.

In practice: a hedge checked once and a hedge checked three times carry different meaning, even though both returned nothing. The record of searching changes where to search next.

Missing piece 2: how to allocate limited time

An owner's time is finite. In what order, and where, should it be spent to maximise the chance of finding the cat?

This was solved seventy years ago. Koopman's theory of search (1956-57) and Stone's Theory of Optimal Search (1975) are the mathematical basis of maritime search and rescue, and have located sunken vessels and missing people.

The idea is simple. Work first where probability of presence multiplied by probability of detection is highest. A place where the cat is likely but could never be seen ranks lower than it looks.

Cat search has exactly this shape. The cat papers supply where it probably is. Search theory supplies the order in which to look. Neither is sufficient alone.

How many people does this concern?

Weiss, Slater & Lord 2012 (Animals 2:301-315), a random telephone survey, reports that 12-18% of cats go missing at least once in five years (dogs: 11-16%).

Because the sample is random rather than drawn from people who contacted a lost-pet service, the bias is smaller than in most published figures.

Do not corner the cat

Cats are strongly stressed by unfamiliar environments and unfamiliar people. The Cat Stress Score of Kessler & Turner 1997, and a body of later work on hiding places, shows that the ability to hide reduces that stress.

A search should therefore be quiet: no surrounding the area with a crowd, no sustained calling, no unnecessary noise around the escape point. Leaving a quiet route home works better than pursuit.

What we are building is not a capture tool. It is a tool that helps a cat come back by itself.

Said plainly

There are only three cat papers here, and some of them report nothing beyond medians and percentiles. You cannot take a paper that gives a median and a 75th percentile, draw a smooth probability curve, and present the curve as the paper's result.

Every number above is as the original reports it. Anything we estimate on top will be marked as ours. Where the literature ends and the KUON model begins is the line that matters most in this field, and we do not intend to blur it.