Safecracker Meets Safecracker

I learned to pick locks from a guy named Leo Lavatelli. It turns out that picking ordinary tumbler locks—like Yale locks—is easy. You try to turn the lock by putting a screwdriver in the hole (you have to push from the side in order to leave the hole open). It doesn’t turn because there are some pins inside which have to be lifted to just the right height (by the key). Because it is not made perfectly, the lock is held more by one pin than the others. Now, if you push a little wire gadget—maybe a paper clip with a slight bump at the end—and jiggle it back and forth inside the lock, you’ll eventually push that one pin that’s doing the most holding, up to the right height. The lock gives, just a little bit, so the first pin stays up—it’s caught on the edge. Now most of the load is held by another pin, and you repeat the same random process for a few more minutes, until all the pins are pushed up.

What often happens is that the screwdriver will slip and you hear tic-tic-tic, and it makes you mad. There are little springs that push the pins back down when a key is removed, and you can hear them click when you let go of the screwdriver. (Sometimes you intentionally let go of the screwdriver to see if you’re getting anywhere—you might be pushing the wrong way, for instance.) The process is something like Sisyphus: you’re always falling back downhill.

It’s a simple process, but practice helps a lot. You learn how hard to push on things—hard enough so the pins will stay up, but not so hard that they won’t go up in the first place. What is not really appreciated by most people is that they’re perpetually locking themselves in with locks everywhere, and it’s not very hard to pick them.

When we started to work on the atomic bomb project at Los Alamos, everything was in such a hurry that it wasn’t really ready. All the secrets of the project—everything about the atomic bomb—were kept in filing cabinets which, if they had locks at all, were locked with padlocks which had maybe only three pins: they were as easy as pie to open.

To improve security the shop ouffitted every filing cabinet with a long rod that went down through the handles of the drawers and that was fastened by a padlock.

Some guy said to me, “Look at this new thing the shop put on—can you open the cabinet now?”

I looked at the back of the cabinet and saw that the drawers didn’t have a solid bottom. There was a slot with a wire rod in each one that held a slidable piece (which holds the papers up inside the drawer). I poked in from the back, slid the piece back, and began pulling the papers out through the slot. “Look!” I said. “I don’t even have to pick the lock.”

Los Alamos was a very cooperative place, and we felt it our responsibility to point out things that should be improved. I’d keep complaining that the stuff was unsafe, and although everybody thought it was safe because there were steel rods and padlocks, it didn’t mean a damn thing.

To demonstrate that the locks meant nothing, whenever I wanted somebody’s report and they weren’t around, I’d just go in their office, open the filing cabinet, and take it out. When I was finished I would give it back to the guy: “Thanks for your report.”

“Where’d you get it?”

“Out of your filing cabinet.”

“But I locked it!”

“I know you locked it. The locks are no good.”

Finally some filing cabinets came which had combination locks on them made by the Mosler Safe Company. They had three drawers. Pulling the top drawer out would release the other drawers by a catch. The top drawer was opened by turning a combination wheel to the left, right, and left for the combination, and then right to number ten, which would draw back a bolt inside. The whole filing cabinet could he locked by closing the bottom drawers first, then the top drawer, and spinning the combination wheel away from number ten, which pushed up the bolt.

These new filing cabinets were an immediate challenge, naturally. I love puzzles. One guy tries to make something to keep another guy out; there must be a way to beat it!

I had first to understand how the lock worked, so I took apart the one in my office. The way it worked is this: There are three discs on a single shaft, one behind the other; each has a notch in a different place. The idea is to line up the notches so that when you turn the wheel to ten, the little friction drive will draw the bolt down into the slot generated by the notches of the three discs.

Now, to turn the discs, there’s a pin sticking out from the back of the combination wheel, and a pin sticking up from the first disc at the same radius. Within one turn of the combination wheel, you’ve picked up the first disc.

On the back of the first disc there’s a pin at the same radius as a pin on the front of the second disc, so by the time you’ve spun the combination wheel around twice, you’ve picked up the second disc as well.

Keep turning the wheel, and a pin on the back of the second disc will catch a pin on the front of the third disc, which you now set into the proper position with the first number of the combination.

Now you have to turn the combination wheel the other way one full turn to catch the second disc from the other side, and then continue to the second number of the combination to set the second disc.

Again you reverse direction and set the first disc to its proper place. Now the notches are lined up, and by turning the wheel to ten, you open the cabinet.

Well, I struggled, and I couldn’t get anywhere. I bought a couple of safecracker books, but they were all the same. In the beginning of the book there are some stories of the fantastic achievements of the safecracker, such as the woman caught in a meat refrigerator who is freezing to death, but the safecracker, hanging upside down, opens it in two minutes. Or there are some precious furs or gold bullion under water, down in the sea, and the safecracker dives down and opens the chest.

In the second part of the book, they tell you how to crack a safe. There are all kinds of ninny-pinny, dopey things, like “It might be a good idea to try a date for the combination, because lots of people like to use dates.” Or “Think of the psychology of the owner of the safe, and what he might use for the combination.” And “The secretary is often worried that she might forget the combination of the safe, so she might write it down in one of the following places—along the edge of her desk drawer, on a list of names and addresses … and so on.

They did tell me something sensible about how to open ordinary safes, and it’s easy to understand. Ordinary safes have an extra handle, so if you push down on the handle while you’re turning the combination wheel, things being unequal (as with locks), the force of the handle trying to push the bolt down into the notches (which are not lined up) is held up more by one disc than another. When the notch on that disc comes under the bolt, there’s a tiny click that you can hear with a stethoscope, or a slight decrease in friction that you can feel (you don’t have to sandpaper your fingertips), and you know, “There’s a number!”

You don’t know whether it’s the first, second, or third number, but you can get a pretty good idea of that by finding out how many times you have to turn the wheel the other way to hear the same click again. If it’s a little less than once, it’s the first disc; if it’s a little less than twice, it’s the second disc (you have to make a correction for the thickness of the pins).

This useful trick only works on ordinary safes, which have the extra handle, so I was stymied.

I tried all kinds of subsidiary tricks with the cabinets, such as finding out how to release the latches on the lower drawers, without opening the top drawer, by taking off a screw in front and poking around with a piece of hanger wire.

I tried spinning the combination wheel very rapidly and then going to ten, thus putting a little friction on, which I hoped would stop a disc at the right point in some manner. I tried all kinds of things. I was desperate.

I also did a certain amount of systematic study. For instance, a typical combination was 69-32-21. How far off could a number be when you’re opening the safe? If the number was 69, would 68 work? Would 67 work? On the particular locks we had, the answer was yes for both, but 66 wouldn’t work. You could he off by two in either direction. That meant you only had to try one out of five numbers, so you could try zero, five, ten, fifteen, and so on. With twenty such numbers on a wheel of 100, that was 8000 possibilities instead of the 1,000,000 you would get if you had to try every single number.

Now the question was, how long would it take me to try the 8000 combinations? Suppose I’ve got the first two numbers right of a combination I’m trying to get. Say the numbers are 69-32, but I don’t know it—I’ve got them as 70-30. Now I can try the twenty possible third numbers without having to set up the first two numbers each time. Now let’s suppose I have only the first number of the combination right. After trying the twenty numbers on the third disc, I move the second wheel only a little bit, and then do another twenty numbers on the third wheel.

I practiced all the time on my own safe so I could do this process as fast as I could and not get lost in my mind as to which number I was pushing and mess up the first number. Like a guy who practices sleight of hand, I got it down to an absolute rhythm so I could try the 400 possible back numbers in less than half an hour. That meant I could open a safe in a maximum of eight hours—with an average time of four hours.

There was another guy there at Los Alamos named Staley who was also interested in locks. We talked about it from time to time, but we weren’t getting anywhere much. After I got this idea how to open a safe in an average time of four hours, I wanted to show Staley how to do it, so I went into a guy’s office over in the computing department and asked, “Do you mind if I use your safe? I’d like to show Staley something.”

Meanwhile some guys in the computing department came around and one of them said, “Hey, everybody; Feynman’s gonna show Staley how to open a safe, ha, ha, ha!” I wasn’t going to actually open the safe; I was just going to show Staley this way of quickly trying the back two numbers without losing your place and having to set up the first number again.

I began. “Let’s suppose that the first number is forty, and we’re trying fifteen for the second number. We go back and forth, ten; back five more and forth, ten; and so on. Now we’ve tried all the possible third numbers. Now we try twenty for the second number: we go back and forth, ten; back five more and forth, ten; back five more and forth, CLICK!” My jaw dropped: the first and second numbers happened to be right!

Nobody saw my expression because my back was towards them. Staley looked very surprised, but both of us caught on Very quickly as to what happened, so I pulled the top drawer out with a flourish and said, “And there you are!”

Staley said, “I see what you mean; it’s a very good scheme”—and we walked out. Everybody was amazed. It was complete luck. Now I really had a reputation for opening safes.

It took me about a year and a half to get that far (of course, I was working on the bomb, too!) but I figured that I had the safes beaten, in the sense that if there was a real difficulty—if somebody was lost, or dead, and nobody else knew the combination but the stuff in the filing cabinet was needed—I could open it. After reading what preposterous things the safecrackers claimed, I thought that was a rather respectable accomplishment.

We had no entertainment there at Los Alamos, and we had to amuse ourselves somehow, so fiddling with the Mosler lock on my filing cabinet was one of my entertainments. One day I made an interesting observation: When the lock is opened and the drawer has been pulled out and the wheel is left on ten (which is what people do when they’ve opened their filing cabinet and are taking papers out of it), the bolt is still down. Now what does that mean, the bolt is still down? It means the bolt is in the slot made by the three discs, which are still properly lined up. Ahhhh!

Now, if I turn the wheel away from ten a little bit, the bolt comes up; if I immediately go back to ten, the bolt goes back down again, because I haven’t yet disturbed the slot. If I keep going away from ten in steps of five, at some point the bolt won’t go back down when I go back to ten: the slot has just been disturbed. The number just before, which still let the bolt go down, is the last number of the combination!

I realized that I could do the same thing to find the second number: As soon as I know the last number, I can turn the wheel around the other way and again, in lumps of five, push the second disc bit by bit until the bolt doesn’t go down. The number just before would be the second number.

If I were very patient I would he able to pick up all three numbers that way, but the amount of work involved in picking up the first number of the combination by this elaborate scheme would be much more than just trying the twenty possible first numbers with the other two numbers that you already know, when the filing cabinet is closed.

I practiced and I practiced until I could get the last two numbers off an open filing cabinet, hardly looking at the dial. Then, when I’d be in some guy’s office discussing some physics problem, I’d lean against his opened filing cabinet, and just like a guy who’s jiggling keys absent-mindedly while he’s talking, I’d just wobble the dial back and forth, back and forth. Sometimes I’d put my finger on the bolt so I wouldn’t have to look to see if it’s coming up. In this way I picked off the last two numbers of various filing cabinets. When I got back to my office I would write the two numbers down on a piece of paper that I kept inside the lock of my filing cabinet. I took the lock apart each time to get the paper—I thought that was a very safe place for them.

After a while my reputation began to sail, because things like this would happen: Somebody would say, “Hey, Feynman! Christy’s out of town and we need a document from his safe—can you open it?”

If it was a safe I knew I didn’t have the last two numbers of, I would simply say, “I’m sorry, but I can’t do it now; I’ve got this work that I have to do.” Otherwise, I would say, “Yeah, but I gotta get my tools.” I didn’t need any tools, but I’d go back to my office, open my filing cabinet, and look at my little piece of paper: “Christy—35, 60.” Then I’d get a screwdriver and go over to Christy’s office and close the door behind me. Obviously not everybody is supposed to be allowed to know how to do this!

I’d he in there alone and I’d open the safe in a few minutes. All I had to do was try the first number at most twenty times, then sit around, reading a magazine or something, for fifteen or twenty minutes. There was no use trying to make it look too easy; somebody would figure out there was a trick to it! After a while I’d open the door and say, “It’s open.”

People thought I was opening the safes from scratch. Now I could maintain the idea, which began with that accident with Staley, that I could open safes cold. Nobody figured out that I was picking the last two numbers off their safes, even though—perhaps because—I was doing it all the time, like a card sharp walking around all the time with a deck of cards,

I often went to Oak Ridge to check up on the safety of the uranium plant. Everything was always in a hurry because it was wartime, and one time I had to go there on a weekend. It was Sunday, and we were in this fella’s office—a general, a head or a vice president of some company, a couple of other big muck-a-mucks, and me. We were gathered together to discuss a report that was in the fella’s safe—a secret safe—. when suddenly he realized that he didn’t know the combination. His secretary was the only one who knew it, so he called her home and it turned out she had gone on a picnic up in the hills.

While all this was going on, I asked, “Do you mind if I fiddle with the safe?”

“Ha, ha, ha—not at all!” So I went over to the safe and started to fool around.

They began to discuss how they could get a car to try to find the secretary, and the guy was getting more and more embarrassed because he had all these people waiting and he was such a jackass he didn’t know how to open his own safe. Everybody was all tense and getting mad at him, when CLICK!—the safe opened.

In 10 minutes I had opened the safe that contained all the secret documents about the plant. They were astonished. The safes were apparently not very safe. It was a terrible shock: All this “eyes only” stuff, top secret, locked in this wonderful secret safe, and this guy opens it in ten minutes!

Of course I was able to open the safe because of my perpetual habit of taking the last two numbers off. While in Oak Ridge the month before, I was in the same office when the safe was open and I took the numbers off in an absentminded way—I was always practicing my obsession. Although I hadn’t written them down, I was able to vaguely remember what they were. First I tried 40-15, then 15-40, but neither of those worked. Then I tried 10-45 with all the first numbers, and it opened.

A similar thing happened on another weekend when I was visiting Oak Ridge. I had written a report that had to be OKed by a colonel, and it was in his safe. Everybody else keeps documents in filing cabinets like the ones at Los Alamos, but he was a colonel, so he had a much fancier, two-door safe with big handles that pull four ѕ-inch-thick steel bolts from the frame. The great brass doors swung open and he took out my report to read.

Not having had an opportunity to see any really good safes, I said to him, “Would you mind, while you’re reading my report, if I looked at your safe?”

“Go right ahead,” he said, convinced that there was nothing I could do.