How a dog reads the street with its nose

What the dogs did#

In that 2005 study, the six dogs, two German shepherds and four Labrador and golden retriever crosses, were led at right angles to the middle of a trail someone had walked an hour earlier, across a row of carpet squares. With 21 footprints to sniff they chose the right way 57 times out of 60. With five, 55 out of 60. With three, 28 out of 60, no better than tossing a coin. Four was never tried.

When the squares were shuffled, each footprint still pointing the way it had been walked, the dogs fell to chance: 29 out of 60. When the walker wore overshoes that kept their smell in, the same: 30 out of 60. The dogs were reading the smell, and its order.

On an airfield near Oslo, four trained German shepherds were sent from about five metres away, at right angles to human tracks three minutes old on concrete and twenty minutes old on grass. A microphone above each dog’s nose counted the sniffs: about six a second. On average the dogs slowed down over about three footprints, laid about a second apart, and took around five seconds to decide. All 20 tests ended with the dog following the track the right way, sometimes after a wrong first turn.

And when two tracking dogs were set on the tracks of a person who had walked backwards, they still went, more often than chance, the way the person had moved, not the way the toes pointed.

How they do it is still open. At the carpet walker’s pace, five footprints span about two seconds, and three about one; the authors of the carpet experiment estimated that the smell of a footprint needs one to two seconds to change enough to be read. On an hour-old trail, two seconds is about one part in 1,800 of its age. The researchers assume the dogs compare how strongly neighbouring footprints smell; the carpet study also considers the opposite signal, the growing smell of what the walker’s smell breaks down into. Neither study measured either.

Six sniffs a second#

A sniff is not a breath. Seven dogs wearing a muzzle that measured their airflow each sniffed new smells four to seven times a second; the shepherds on the airfield, about six.

The trick is in the breath out. A dog breathes out downwards and to the sides, not straight ahead, so it doesn’t blow away the smell it is about to take in. In a computer model of one dog’s nose, and in a 3D-printed copy, those jets stir the air and help draw more of it towards the nose. The computer model also suggests that each nostril draws air from its own small patch of ground, which could let a dog compare left and right as two ears compare sounds. That part is still a hypothesis.

Let them read#

In a study published in 2019, two researchers split 20 pet dogs into two groups. For two weeks, half practised nosework at home, finding treats hidden in boxes; the other half practised heelwork, walking at the owner’s side. Before and after, each dog met a bowl placed halfway between where food had always been and where it never was.

After the nosework, dogs went to the uncertain bowl faster, in 3.4 seconds on average instead of 4.7. The authors read it as optimism. The heelwork dogs also got faster, from 5.3 to 4.4 seconds, but not by enough to rule out chance. The groups were small, ten dogs each.

The study did not test walks. Its authors’ own advice was to give dogs more time for their noses, “such as natural sniffing during walks”.

So tomorrow morning, when the lead goes tight at a lamp post, your dog is reading the street. Give it the time.