San Francisco Earthquake Safe

This one is less boring for the public than the others. Hell, I don't know; maybe it's more boring. But  warning: it's pretty long. It's about a job I did a couple of years ago. It involved opening a safe that was in a downtown building that got destroyed by the devastating fires that raged as result of the 1906 earthquake that destroyed San Francisco. This story originally ran in the industry newsletter I publish.

I've been hearing and reading earthquake history ever since I moved to this area, but never guessed I'd have even a small part in recovering any of the history on it.

I got a call from a construction foreman. While remodeling a downtown San Francisco building, his workers had unearthed a large old safe under a basement floor. The safe had been buried since the 1906 earthquake that destroyed most of downtown San Francisco.


Though he wasn’t clear on how the main outer double door safe got opened, it was gone. All that remained was a small, heavy chest that was found inside the big one. Locked shut.


He couldn’t describe the chest well enough to give me an inkling. I told him to email photos. Pictures beat struggling to understand callers’ safe descriptions, and are better than traveling to hear “Thanks for looking.”


The photos he sent were of a MacNeale & Urban pressure chest, the type commonly installed in large fire safes. A year earlier I’d opened an identical “found” chest (see below). That one had some rust and corrosion, but after having been buried for a century, this one gave new meaning to the term “rust bucket.”



Feasible nonetheless, I thought, so I gave an open-only price. He thanked me, said he might call back soon. He did, and we scheduled. On opening day, though, he postponed.


According to him, an archaeological consulting company was now part of the equation. Invoking higher authority (the building and by extension the safe’s owners), the archaeologists immediately forbade anyone to do anything to the rusted chest. The foreman said I’d probably hear from one of them.


Archaeologists. . . ? As far as I knew, this was just another “found” safe, an old, rusted mess.


Soon after that, a man named Joe called, saying he was the archaeologist who would be handling the safe. I asked him how the job had devolved from me dealing with construction roustabouts to scientists of antiquity.


Joe explained: The building under which the safe had been buried was built immediately after the San Francisco earthquake of 1906. That monster quake knocked down plenty of buildings, but the real devastation came afterward.


Fires broke out all over The City and burned out of control for days. The building that housed the safe burned and collapsed. The MacNeale & Urban had apparently done a classic crash-through to the basement when the floor under it burned out.


The weeks, months, and years after the `06 quake and fires were consumed by recovery work and rebuilding.  Many businesses paid drayage companies to haul their charred safes from the ruins in hopes of recovering contents.


The MacNeale & Urban in question had obviously been left where it landed. The current building was built over the ruins of the first one. 


Joe and I agreed there were various possible explanations for why the “found” safe never got recovered:


a) The safe’s owner might have had a chance between the early morning quake and the outbreak of the fires to have reached his offices and emptied outer and inner safes of valuables. In that case, nothing left to recover.


b) The owner could have been killed in the quake, or in the fires, epidemics, and civil unrest that followed.


c) The owner could have survived, but might have concluded that safe and contents were a total loss, and decided to leave it where it had fallen.


Over a century later, here it was, exhumed, rusted, corroded, and closed. The foreman said that the outer safe was found open and empty, and that the inner safe was loose inside. I’m not sure if everyone believed that. I wasn’t there when the outer safe was found, but the inner safe appeared to have been forcibly torn from its wall moorings inside the outer safe. There were still sheets of the outer safe’s wall attached to the smaller safe’s sides. Falling doesn’t do that.


The safe had in all likelihood been owned by a business in the building. Exactly which business wasn’t clear. Joe also told me the outer safe had an owner’s name painted on it and researching the name had unearthed some intriguing information: The original safe owner was Barrett & Sherwood, Watch & Chronometer Makers.


Throughout the last half of the nineteenth century Barrett & Sherwood was one of the very few (if not the only) places on the Pacific coast where a ship’s master could have his vessel’s chronometer repaired and calibrated. In the days of celestial navigation, an inaccurate chronometer put a ship at risk of being off-course by anywhere from miles to hundreds of miles.


Barrett & Sherwood erected an observatory with a tower that was visible from the deck of every ship in the harbor. Every day, at what they represented as “nearly noon,” a ball would be dropped from Barrett &Sherwood’s tower. The exact second of the drop was a secret, answer available by subscription. (Perhaps explaining the origins of the New Year’s Eve ball-drop at Times Square?)


It worked like this: A ship’s master would watch for the noon ball drop. Reading from his own chronometer, he would note exactly how many seconds before or after noon the drop occurred. After accumulating at least three days’ notations, paid subscribers would be notified of the precise times of the drops for the noted days. By averaging, then, a ship’s master could calculate the error of his own chronometer against the (presumably correct) times  of Barrett & Sherwood’s main chronometer, then factor that in his navigational calculations at sea.


Why not just bring the chronometer ashore for hand calibration and adjustment on Barrett & Sherwood’s workbench? Many ship’s masters did so, but there were at least two justifications for using the observatory method.


Ship’s chronometers had to be the most accurate timepieces money could buy. And rightly so, considering that seamen’s lives and livelihoods depended on reliable navigation, which was impossible without accurate chronometers. Detaching a ship’s chronometer from its mount, carrying it from ship to shore by dinghy, then by foot or carriage down crowded unpaved streets to a workshop was to risk accidental damage. One stumble, one fumble, or a bad bouncing over a rutted street, and a chronometer could be irreparably broken. Once calibrated, there was the same risk on the return trip to the ship. Some ships’ masters thought it too risky.


Second, a ship chronometer was the high technology of the era. Masters of foreign ships needed to do their trading and provisioning in ports, but many were uneasy about allowing their most sensitive instrumentation to be out of their sight, in the hands of others. All of which made observatory calibration via the ball-drop method preferable for some shipmasters.


While that was interesting information to learn, it still wasn’t clear who owned and used the safe in question when the 1906 earthquake occurred.


Sketchy records showed that Barrett & Sherwood operated in San Francisco from 1852 to 1902, and that their observatory tower was at 161 Clay St. The safe was found under a building located at 1 Kearny St., several blocks away. Furthermore, The original 1 Kearny St. building was built in 1902.


It’s possible that the safe was bought when (and if) Barrett & Sherwood supposedly dissolved in 1902. Maybe the company assets were sold off. The safe might have been hauled from its original home at 161 Clay to 1 Kearny and used by a new owner until 1906.  Another possibility was that the Barrett & Sherwood Observatory on Clay St. was an observatory only, and that the company conducted business at 1 Kearny.


Nobody knew for sure, but regardless of who last owned the safe, it was thought that because it was never hauled out of the ruins, the inner chest might still hold contents.


I understood then. The San Francisco Earthquake of 1906 has been studied, researched, written about, commemorated, and depicted ad nauseam in the years since. Anything related to it still commands high interest. The reasoning, then, was that anything found in the safe had potential historic value. Hence the archaeologists.


Joe asked some questions about the inner safe and my plans for opening it. He wanted to know if I knew the maker’s name, was I familiar with that kind of safe, did I think I could get it open without damaging whatever might be inside, and what method(s) I would be using to do the opening.


After hearing my answers and assurances, he said I’d probably hear back, and we hung up. A week later a lady from his company called with similar questions. She asked for pricing, saying the job had to be witnessed, and that the swinging of the door would need to be on video.


As interesting as it all sounded, I realized this job had potential for turning into a festival of aggravation.  I envisioned a variety of delays and annoyances: Getting in and out of the building and to the safe, multiple parking meter feedings, Q&A sessions with whoever was present, people in my way, and the inevitable impatient onlookers who start asking “how much longer do you think it will take?” five minutes after a job starts.


The best defense is to levy surcharges for them. Aggravate the safecracker, aggravate the price. So I made stipulations, with dollar costs for noncompliance:


I insisted on a prep visit, during which I would remove major surface rust, then apply copious amounts of penetrating fluid to doorjamb, handle arbor, dial spindle, hinges, and pressure cams.


For that I needed them to position the safe with door facing the ceiling, and included that as a condition.


They could bring as many onlookers/witnesses as they liked, I said, but all present had to sit quietly, out of my immediate work area. I recommended bringing something to read.


Last, payment upon job completion, definitely not contingent upon whether or not they found contents. I like science and history just fine, but eating regularly ranks right up there on my list of priorities.


There was a communications lull of a few weeks. Then the lady emailed to acknowledge and say they wanted to do it the next week. We scheduled.


I visited on a Thursday morning and met Joe the archaeologist. He wanted to be present for every phase. We went up to a secured empty floor of the building, where the safe sat on a dolly, door facing the ceiling, per my stipulation.


Basics first: I scraped, wire-brushed, and blew off the loose rust and scale, then started flooding the jamb with rust-penetrating PB Blaster. Next I flooded the handle arbor base and tested for movement on dial and bolt control handle. The dial was frozen solid. Light tapping on the bolt control handle with a dead blow hammer seemed to be getting a teensy bit of movement, but it was actually the threaded end of the cast grip bending a little just before it snapped off.


Drat . . . I’d half-expected that. Besides, I’d warned Joe there would be damage. No matter; there was still a big sturdy arbor end to clamp a vise grip onto if need be.


I managed to remove the screwed-on lettered dial intact. However, a large vise grip on the spindle couldn’t get even fractional movement.


Judging from the safe’s exterior and its history, I’d have been surprised not to find the spindle rust-frozen. The spindle was steel,  very closely-fitted when new, probably punch-and pull-resistant and therefore trapped in the steel door laminations.


Based on having opened the other identical MacNeale & Urban a year before, I knew this door was comprised of five screwed-together laminations of half-inch steel, with laminations alternating from mild, hardened, mild, hardened, and finally mild.


I left the dial ring in place so it could act as a tinker’s dam, then flooded it, creating a little lake of penetrant around the spindle end. After flooding the handle arbor some more, I used the remainder on the jamb edges.


I wanted to return Friday or Saturday for one more flooding session, but was told there wouldn’t be anyone to give me access. We agreed to meet and begin in earnest first thing Monday morning, then left.


On Monday Joe and the building manager met me outside and we went upstairs. There was still no hint of movement on handle arbor or dial, but the pressure handle was movable. No visible in-out door play, but there’s little or no play when a pressure door is in excellent condition.


Because I thought I knew which lock was inside, my plan was to make a first hole into the lock at the drop-in point. It wasn’t like I expected to probe wheel gates into place. That would have been wishful thinking with the spindle and driver immobilized. Even if that were somehow possible, an immovable driver would preclude bolt retraction.


My reason for doing it was just to see if the wheels and especially the lever had any wiggle. That answer would dictate my next step.


If wheels and lever had movement, I planned to drill for the screw attaching the lever to the lock bolt and probe the bolt back. If I found the lever immovable, I’d have to assume the lock bolt was, too. If that turned out to be the case, my plan was drill to the lock bolt just outside the case, then punch it out of the handle cam’s way.


Once the lock was neutralized, the grunt work could begin: Side-drill to each of the four door bolts’ ends and drive them back.


Drilling to the drop area wasn’t difficult with a lever rig and bits. While drilling, I got further evidence of through-and-through rusting. The bit popped into an  air gap between two of the laminations, and as it dug into the next lamination a little puff of rust dust shot out around the flutes.


After drilling to the bronze of the lock case, I switched to a high speed bit to minimize burring at breakthrough.  When the hole was finished I wasn’t seeing what I expected to see. A piece of metal angled upward across the left third of the entry hole. It wiggled a little when probed, but not much. If there were any top edges of driver or combination wheels to be seen, this piece was blocking my view. Whatever was to its right wasn’t movable, though.


After puzzling over this, I decided to drill another hole where a diagram I had showed the lever screw. I figured on drilling it out and probing to retract the lock bolt. The fact that something in that lock could still wiggle after a century made me feel that was still feasible.


That second hole took about the same time and effort; unremarkable.


A peek into the lock did not, however, show me the lever screw. More scoping and probing; what I’d come in on also had a little wiggle to it. But what?


One parking-meter-feeding break later  I thought I had it figured out: The damned lock bolt was already retracted. Extensive peeking with first a 90° then a 70° scope seemed to verify this. The thing that was angled across my drop-in hole was the arm of the lever, pulled back to where it should be at bolt retraction. Fence and lever nose naturally weren’t visible, because  bolt retraction moves both parts out of the drop-in hole’s line of view.


Two possibilities: When the entire safe fell to the basement, then, the inner chest had apparently had its door closed and pressured inward for door bolt clearance and extension, but the combination lock bolt hadn’t been extended.


Or, perhaps the last person to close the inner door had left the safe on “day lock” or “day hitch.” In other words, combination wheels, gates still dialed to alignment and the dial just turned the short distance necessary to extend the lock bolt. People disliked redialing as much then as now. It was conceivable that the major jostling that had to have occurred as the entire safe fell and landed in the basement had allowed the substantial lock bolt to retract by inertia alone.


I didn’t feel too badly about the time spent drilling and scoping just to learn this. The rust-frozen dial and spindle hadn’t lent themselves to any diagnostics. “So,” I told myself, “all I need to do next is figure out measurements to get me on the four bolt ends, mark, drill for them, and drive them back.” I scaled measurements from the photos of the earlier MacNeale & Urban, applied them to the safe sides, and drilled.


The side drilling went faster than the door drilling, but I wasn’t racing anybody. The first drilled hole came in a little low. I opened it up to a larger diameter in order to keep my punch as much on the bolt end as possible. Using a corrected measurement, I drilled the other three holes without incident.


Before starting to punch, I trained the snorkel tube from another can of PB Blaster on each of the bolts and gave them all a bath. I also locked a large vise grip on the end of the handle arbor to give me a visual reference on any bolt/cam/handle movement my effort might be having.


The first punching interlude was on the opening side of the door. The vise grip I’d locked onto the handle arbor never wiggled. Enlarging the holes allowed bigger punches, but still nothing. Thinking that this situation might be helped by a more balanced distribution of punching impacts, I took my business over to the hinge-side door bolts’ ends.


Before starting to punch I sprayed more penetrant on the bolts where they disappeared into their guide rails. I wasn’t counting on the penetrating fluid to work any miracles, though. When looking over my photos of the other M&U chest I’d opened, I’d counted at least a dozen areas where the bolts and their guide rails could have rusted and corroded together. However, I reasoned that it couldn’t hurt either.


My punching was vigorous to say the least. It didn’t seem to be doing a damned thing, so after a good series of alternating smacks on the hinge side bolt ends I went back and balanced the effort on the opening side.


The job was starting to suck in a big way. No apparent progress on the opening side, so back to the hinge side. More alternating smacks, then I finally noticed a little of that “squishy” (for lack of a better word) feel one gets when there’s some give to what you’re punching. Instead of stopping and looking again I gave several more heavy shots to each bolt. Then I looked.


The hinge side bolts were almost flush to their guide rails. Looking good there, but when I had a look at the opening side bolts there was no change. I gave them a few more smacks, but still no change. I was disappointed to say the least. Ordinarily all the bolts on that door design move together.


I smacked some more, but it was closing in on downtown towaway time. The City’s meter maids and their flying tow truck monkeys have that part of every weekday too well choreographed to trust luck. I made a plan with the archaeologist to resume work the following morning.


In the interim, I thought it over: All the bolts were yoked; designed to move together. If the two on the hinge side had moved independently of the two on the opening side, whatever connection was yoking the two sets of bolts had to have been broken by my hinge-side punching.


Also, despite the lost connectivity, the fact that the first two bolts moved suggested that whatever rusting and corrosion existed wasn’t severe enough to completely preclude side-punching.


If so, what was blocking the remaining two bolts’ movement? I’d already seen that the combination lock bolt was retracted, right? But had I really seen clear airspace where the lock bolt sits when doing its job?


No. My two holes into the lock had entered the body, not the area outside the body. Verify clearance, right?


The way I figured it, I had a choice of drilling to get to that space where an extended lock bolt would be, or just assume part of the lock was still extended and drill for a very simple and basic cam punching exercise.


When I returned the next day I chose the first option. My reasoning was that as simple as it sounded, punching the relatively beefy cam might create a further handle shaft bind, and make the movement I needed to impart to it even more difficult to achieve.


Better to get a hole into the lock bolt extension area first and have a look. If the bolt was still extended I could smack it off through that hole. If not, I could look up, see the cam (I hoped) and figure out my next move.


That was how I started the day, then: A hole drilled to the space where the lock bolt would be when extended.


Once it was made, I put a scope in there and got the picture: The handle cam was indeed visible, and its movement was blocked by about 3/32” of the “retracted” lock bolt. It’s always so simple and obvious when you know the answer, isn’t it?


A little more work enlarging the hole to the lock bolt area allowed me to wedge it that little bit further into the lock body. When I could visually verify the clearance I went back to the side and started whaling on the punches again. Even with the clearance the handle cam needed it still took a lot more smacks to move the two bolts.


When the big vise grip I’d locked onto the handle arbor finally showed fractional movement, I began alternating between bolt-punching and smacking the vise like a cheater bar. Finally I looked into the side holes and the last two bolts were beaten back.


At that point it was just a matter of swinging the door. I told Joe the archaeologist. He called the building manager, who needed to collect the building owner and some others, then Joe called his office and had them send over the videographer. We broke for coffee while the party assembled, then returned.


The videographer wanted the door to remain facing the ceiling so the light would be better, so we left the safe on the dolly. Working against gravity like that required locking vise grips onto the pressure bar, because I really didn’t know what to expect from the cast pressure handle.


Once the door started outward I had to  do some wedging and prying to overcome the hinge side’s tendency to want to stay behind and cause more of a jam (remember, this door had center-pivot crane hinge, so it wanted to rock as it swung). Once I’d wedged the tips of a couple of pry bars securely under the edge, I paused and told everyone to keep their hands at their sides if they saw my prying tools losing their grip. If that happened, I told them, I’d simply wedge and pry again. I didn’t want any valiant grabs by any onlooker who wanted to help if they saw the door falling shut.


It did slip back a couple of times before I had a good wedge, but I finally got it to come out squarely and swung it upward. The hinge rust and corrosion was so pronounced that getting the door to 90° open took some muscle, and then it stood upright without support.


More cautions to all present that no hands should be reaching into the safe interior while the door was like that, then everyone moved in closer to peer at the . . . CONTENTS!



I don’t usually do much rubbernecking at customers’ safe contents, but this time was different because of the historical slant to the job. The onlookers were an orderly bunch, waiting obediently a few steps back until I gathered tools before motioning them forward.


Joe removed each item from the chest as if on an archaeological dig, using gloves, a long forceps, and extreme care. Needless to say, it went very slowly. I had to leave before  everything was laid out on the acid-free paper Joe had spread across a nearby desktop. He told me about the items of interest later and sent me the contents photo shown on this page.


Joe described the singed papers as being like baklava crust, thin, curled, brittle, and extremely fragile. Much of the printed material was illegible, but among the more easily identified items: An old Wabash Railroad train schedule, an old ticket for a train trip to Monterey, and some banking papers. Of more interest were the four gold Krugers, a stickpin, a ring, and a brooch.


Joe is still trying to learn who last owned the safe.

That was one of my more interesting jobs.

Ken Dunckel



Popular posts from this blog

Antique Safe Prices & Values: "How Much Is It Worth?"

How To Dial A Safe Combination

Safe Boobytraps: Tear Gas and Unknown Contents