Every once in a while the army sent down a lieutenant to check on how things were going. Our boss told us that since we were a civilian section, the lieutenant was higher in rank than any of us. “Don’t tell the lieutenant anything,” he said. “Once he begins to think he knows what we’re doing, he’ll be giving us all kinds of orders and screwing everything up.
By that time I was designing some things, but when the lieutenant came by I pretended I didn’t know what I was doing, that I was only following orders.
“What are you doing here, Mr. Feynman?”
“Well, I draw a sequence of lines at successive angles, and then I’m supposed to measure out from the center different distances according to this table, and lay it out.
“Well, what is it?”
“I think it’s a cam.” I had actually designed the thing, but I acted as if somebody had just told me exactly what to do.
The lieutenant couldn’t get any information from anybody and we went happily along, working on this mechanical computer, without any interference.
One day the lieutenant came by and asked us a simple question: “Suppose that the observer is not at the same location as the gunner—how do you handle that?”
We got a terrible shock. We had designed the whole business using polar coordinates, using angles and the radius distance. With X and Y coordinates, it’s easy to correct for a displaced observer. It’s simply a matter of addition or subtraction. But with polar coordinates, it’s a terrible mess!
So it turned out that this lieutenant whom we were trying to keep from telling us anything ended up telling us something very important that we had forgotten in the design of this device: the possibility that the gun and the observing station are not at the same place! It was a big mess to fix it.
Near the end of the summer I was given my first real design job: a machine that would make a continuous curve out of a set of points—one point coming in every fifteen seconds—from a new invention developed in England for tracking airplanes, called “radar.” It was the first time I had ever done any mechanical designing, so I was a little bit frightened.
I went over to one of the other guys and said, “You’re a mechanical engineer; I don’t know how to do any mechanical engineering, and I just got this job.”
“There’s nothin’ to it,” he said. “Look, I’ll show you. There’s two rules you need to know to design these machines. First, the friction in every bearing is so-and-so much, and in every gear junction, so-and-so much. From that, you can figure out how much force you need to drive the thing. Second, when you have a gear ratio, say 2 to 1, and you are wondering whether you should make it 10 to 5 or 24 to 12 or 48 to 24, here’s how to decide: You look in the Boston Gear Catalogue, and select those gears that are in the middle of the list. The ones at the high end have so many teeth they’re hard to make. If they could make gears with even finer teeth, they’d have made the list go even higher. The gears at the low end of the list have so few teeth they break easy. So the best design uses gears from the middle of the list.”
I had a lot of fun designing that machine. By simply selecting the gears from the middle of the list and adding up the little torques with the two numbers he gave me, I could be a mechanical engineer!
The army didn’t want me to go back to Princeton to work on my degree after that summer. They kept giving me this patriotic stuff, and offered a whole project that I could run, if I would stay.
The problem was to design a machine like the other one—what they called a director—but this time I thought the problem was easier, because the gunner would be following behind in another airplane at the same altitude. The gunner would set into my machine his altitude and an estimate of his distance behind the other airplane. My machine would automatically tilt the gun up at the correct angle and set the fuse.
As director of this project, I would be making trips down to Aberdeen to get the firing tables. However, they already had some preliminary data. I noticed that for most of the higher altitudes where these airplanes would be flying, there wasn’t any data. So I called up to find out why there wasn’t any data and it turned out that the fuses they were going to use were not clock fuses, but powder-train fuses, which didn’t work at those altitudes—they fizzled out in the thin air.
I thought I only had to correct the air resistance at different altitudes. Instead, my job was to invent a machine that would make the shell explode at the right moment, when the fuse won’t burn!
I decided that was too hard for me and went back to Princeton.
Testing Bloodhounds
When I was at Los Alamos and would get a little time off, I would often go visit my wife, who was in a hospital in Albuquerque, a few hours away. One time I went to visit her and couldn’t go in right away so I went to the hospital library to read.
I read an article in Science about bloodhounds, and how they could smell so very well. The authors described the various experiments that they did—the bloodhounds could identify which items had been touched by people, and so on—and I began to think: It is very remarkable how good bloodhounds are at smelling, being able to follow trails of people, and so forth, but how good are we, actually?
When the time came that I could visit my wife, I went to see her, and I said, “We’re gonna do an experiment. Those Coke bottles over there (she had a six-pack of empty Coke bottles that she was saving to send out)—now you haven’t touched them in a couple of days, right?”
“That’s right.”
I took the six-pack over to her without touching the bottles, and said, “OK. Now I’ll go out, and you take out one of the bottles, handle it for about two minutes, and then put it back. Then I’ll come in, and try to tell which bottle it was.”
So I went out, and she took out one of the bottles and handled it for quite a while—lots of time, because I’m no bloodhound! According to the article, they could tell if you just touched it.
Then I came back, and it was absolutely obvious! I didn’t even have to smell the damn thing, because, of course, the temperature was different. And it was also obvious from the smell. As soon as you put it up near your face, you could smell it was dampish and warmer. So that experiment didn’t work because it was too obvious.
Then I looked at the bookshelf and said, “Those books you haven’t looked at for a while, right? This time, when I go out, take one book off the shelf, and just open it—that’s all—and close it again; then put it back.”
So I went out again, she took a book, opened it and closed it, and put it back. I came in—and nothing to it! It was easy. You just smell the books. It’s hard to explain, because we’re not used to saying things about it. You put each book up to your nose and sniff a few times, and you can tell. It’s very different. A book that’s been standing there a while has a dry uninteresting kind of smell. But when a hand has touched it, there’s a dampness and a smell that’s very distinct.
We did a few more experiments, and I discovered that while bloodhounds are indeed quite capable, humans are not as incapable as they think they are: it’s just that they carry their nose so high off the ground!
(I’ve noticed that my dog can correctly tell which way I’ve gone in the house, especially if I’m barefoot, by smelling my footprints. So I tried to do that: I crawled around the rug on my hands and knees, sniffing, to see if I could tell the difference between where I walked and where I didn’t, and I found it impossible. So the dog is much better than I am.)
Many years later, when I was first at Caltech, there was a party at Professor Bacher’s house, and there were a lot of people from Caltech. I don’t know how it came up, but I was telling them this story about smelling the bottles and the books. They didn’t believe a word, naturally because they always thought I was a faker. I had to demonstrate it.