Evidence, not adjectives.

We were called out to work a vehicle for another agency. When we got there five or six patrol cars were already on scene. Must be a big deal, I thought. The officer who made the stop asked me to run Tatsa on the pickup and tell him what we found.

I took her to the grass for a break and we approached from the passenger side. She engaged and started to work. Coming past the driver's door she stopped hard, changed direction back toward it, sniffed for a minute, and sat with her nose pointed at the door. I signaled the alert. The officer opened the door and reached in behind the seat almost immediately: a medicine bottle holding thirty crack rocks, individually wrapped in foil and packaged for distribution. The narcotics and a quantity of cash were seized and the driver was arrested.

Then I was told a K9 team had already worked that truck and cleared it. Had I known that before I came, I probably would not have come.

That one stayed with me. I wanted to know why the other dog did not alert, and I have since learned there are a lot of possible reasons. But one of the most common is handler influence — and almost everyone discusses it in one direction only. Handlers are trained not to cue a dog into a false alert. I think influence can just as easily cause a miss. If a dog has become handler-dependent and is waiting on a cue it is used to getting, it can find the odor, never get the cue, and never give the trained final response.

The handler that day was new, and this is a hypothesis on my part, not something I can prove about that dog. But knowing what I know now it is a reasonable explanation, and it is a large part of why I built training strategies to reduce handler influence. We have to guard against influencing a dog into a false alert. We also have to guard against influencing one into a miss.

After Boomer passed away I started training my wife's Border Collie, Lucy, on explosives. She took to it almost immediately, and before long we were training regularly with the bomb squad from a larger department. Those men were professionals who believed training should be as realistic as possible. Their commander built scenarios that forced the team to think like investigators rather than technicians — evaluate the evidence, make a decision, solve the problem under conditions that looked like a real call.

One exercise has stayed with me ever since. A cardboard box had been left at the entrance to a shopping center before opening, reported by patrol after a bomb threat. The center was closed, so the squad had the whole property to work with. They approached the box, elected not to X-ray it, and decided to deploy Lucy to find out whether explosive odor was present.

I gave her the command to search. The box was the only object in the area, so of course it drew her. She worked it thoroughly, every side. I watched for the behavior that always came before her trained final response when she was in odor. It never came. No increase in interest. No change in breathing. No increased focus. No source commitment. She satisfied her curiosity, left the box, and carried on searching the area around it. She had run the problem and reached a conclusion. It just was not the conclusion the humans expected.

After a few minutes I called her back and reported that Lucy found no indication of explosive odor. For reasons I have never fully understood, the decision was made to proceed as though a device was inside anyway. A water cannon was brought up, the stand-off distance calculated, the operators careful and methodical and entirely professional. Then they fired. The box came apart and a life-sized plastic baby doll bounced across the parking lot.

Nobody said anything for a moment. The exercise had never been about destroying a suspicious package. It was about making a sound decision on the available evidence. By that standard the team failed. Lucy did not. She investigated, evaluated, reached a conclusion and moved on. She never hesitated, never guessed, and never changed her answer because experienced people believed something different.

That is what Credo Vestri Canis means to me. Trust your dog.

Boomer was my first explosives dog and he taught me more than any dog since. One thing he did looked genuinely impressive: he could find a single round of ammunition, fast. I ran that demo many times. Pop the magazine out of my Glock, take out a round, toss it into grass or somewhere it was out of sight, send Boomer, and within a few minutes he would be sitting and pointing his nose at it. It never failed.

Then one day another officer and I were at a gathering with a lot of other police there. I had Boomer with me for socialisation. My colleague wanted to show people what the dog could do, so he popped a round out of his Glock and threw it into a grassy area. I sent Boomer and he engaged hard, sniffing, working. I had no reason to think he would not be sitting on it within the minute.

The minutes went by. It became obvious he was not going to find it. That was one of the most humbling things that has happened to me in this work, and it happened in front of a crowd of cops.

I thought about it for a long time afterwards, mostly because I never wanted it to happen again. I knew dogs could find ammunition, so something was wrong, and I suspected the something was me. It was. Eventually it dawned on me what had actually been going on. Boomer was not finding a bullet. He was finding an object that carried the odor of my hands. Every round I had ever thrown had been handled by me first.

That taught me three things I have never let go of. Human odor is enormously strong to a dog. A dog can tell one person's odor from another's. And a training set-up can quietly teach a dog something completely different from what you believe you are teaching it, while producing results that look like success every single time.

A series of arson fires was being set at multi-family apartment complexes. We were called in by another agency to work them — the mayor and chief I worked for always encouraged helping other departments, on the understanding that the favour comes back, and it did on many occasions.

The arsonist always set the fires on the back decks. He had started with dumpster fires and graduated, so to speak, to structures. From the first scene we worked through the last, Midnight found the residue of charcoal lighter fluid. Sometimes it was a pour that ran up the steps and onto the deck. On others it was in the soil under the deck, where the fluid had run between the boards and dripped through to the ground below. She found hot samples at every one of the fires we worked.

What mattered was not any single alert. It was that her alerts kept landing in the same two places — on the deck near the back door, and at the top of the steps — scene after scene. The lab confirmed the samples. That consistency is what turned a run of separate fires into a pattern, and the pattern was a component of tying seven of them together.

Commercial bed bug work is a cost problem before it is a pest problem. Treating an entire auditorium means chemical, labour, and closed doors on a room that only earns money when people are sitting in it. Treating the seats that actually have bugs in them means a fraction of all three.

Our teams worked theaters seat by seat and isolated the infested ones. Individual seats were treated rather than the whole theater, which saved the operators thousands in treatment costs and kept rooms in service. The venues were eventually cleared, and after that only periodic inspection was needed to keep them that way.

That is the case for a dog in commercial detection, and it is an economic one as much as a technical one. A dog that can tell you which twelve seats out of four hundred are a problem is not a novelty. It is the difference between a targeted treatment and shutting the room.