Baldwin adding machine
This Baldwin adder was sold on an American auction website, and the Arithmeum was lucky enough to acquire it. It seems to be very rare - I only know of one other example, which is the printing version that is kept in the Smithsonian.These are some pictures by the seller, to be able to judge the condition it arrived in here.
I’ve made some more pictures of course, also of the insides.
The principle of this machine is very simple - it has a kind of clock mechanism with a spring that is wound by the user. Pressing any key will block the gear wheel after the number of teeth on the key, and the gear will advance the spoke-wheel that protrudes through the top plate of the machine by an equal number of teeth. Then the carriage lock will be pushed up, advancing the carriage to the left for the next digit. In that way numbers can be sequentially added to the register.
Now, what’s wrong with it?
- The lid of the box was somewhat broken and needed glueing and a coat of wax - that is an easy fix.
- The machine was dirty and stiff, as you expect for an old machine
- The bracket holding the carriage onto the calculator wasn’t fully tightened. It turns out that if you fully tighten it, that the carriage no longer sits properly on the machine. It looks like a part could be missing from the front of the carriage as well. That is a bit of a puzzle.
- Something strange is going on with the clearing of the carriage. The rightmost number does not clear to 0, but to 8, and this messes up the clearing in all the rest of the carriage, which should terminate with one long glorious tens’ carry from left all the way to the right, but as that is not happening, all the carries are happening at once, leading some numbers to end up at 0, others at 1, and others still at 2. That should actually not be possible, but for some reason it is anyway. The wheels hook onto a slot in the central axle with a spring loaded hook, exactly like in later Monroes.
So what to do about it then ? I decided to deal with the assembly of the carriage later, and first wanted it function reliably. So the carriage was separated from the machine, the keytops were removed (they screw on) and the whole machine was dunked in the ultrasonic bath in order to remove dust, dirt and old lubricant. After washing three times in demineralised water, the machine went into the oven at 60 degrees for 25 minutes, and came out nice, clean and dry. It was then lubricated with teflon containing lubricant to prevent rust.
The carriage got its release lever bent back into shape and the pins for the release lever and the zeroing knob knocked out, and they were removed.
Then the carriage was systematically taken apart and sewn onto steel wires, for the separate parts to be dunked into the ultrasonic bath as well to clean them.
In these pictures you can see how the centre of the rightmost wheel is twisted with respect to the outer rim.
After drying, reassembly was the reverse of disassembly, and we arrived back at the stage where the first wheel clears to completely the wrong position, screwing everything up.
The fix has to be that the center of the rightmost wheel is turned over two digits. That turned out to be quite feasible - it’s not loose, but it definitely isn’t holding on for dear life either.
It did not, however, fix things. I have learned a lot in the past few days about how the carriage of this little Baldwin adder is supposed to function. So I’ll explain it here in a bit more detail, so that you don’t end up thinking, like me during the darker moments of this restoration, that the thing can never have worked right in the first place. No wonder it was never a commercial success!
First of all, let’s have a look at the wheels. They roll five functions into one. First of all, they need to show numbers, of course. These are marked at the outer edge. Secondly, the wheel has slots between the numbers, by which it engages with the spokewheel that drives it. Thirdly, looking at the wheel from the front, the pegs used for locating it precisely are sticking out on the righthand side.
Then, it has a brass center which shows a slowly increasing curve over about 60% of its circumference - this is for lifting the tens’ carry arms. And finally, this brass center has a spring loaded pawl, which grips in the slot in the central axle, and is aligned with the 0, to be used in clearing.
Now all the metalwork at the back - There are two arms - a carry arm, and a location/locking arm. Looking in the picture below, the locking arm is in the front.
This is the view of the left side of a numeral wheel with its carry and locking mechanism (it is here at 9, so the roller is at the end of the curve, and the tens’ carry arm is pushed all the way back)
The locking arm pivots independently from the tens carry arm, and has a “tail” in the back, that falls into a flat-bottomed slot in the bar at the back of the carriage. When this bar is tilted forward by the clearing unlock thumblever on the left side of the carriage, this tail is pushed down, and the nose of the locking lever is lifted back from between the pins on the wheel to its left.
Behind the locking arm sits the tens carry arm. This interacts with the locking arm in intricate ways. It is spring-loaded to be pushed forward, and has a little roller sticking out to the right side, that follows the rising curve on the inside of the wheel and hence pushes the tens carry arm back against the spring pressure. When pushed back far enough, the hinged nose of the tens carry lever is spring-loaded against the locking lever, and falls down, behind one of the pins on the numeral wheel. When this happens, the tens carry arm is already quite far back, and by a protrusion on its left side (red arrow in picture below), also pushes back the locking arm, out of engagement with the pins on the wheel to its left. This is so that when the wheel to the right shifts from 9 to 0, the roller drops off the curve, the nose of the tens’ carry arm pushes forward on the peg to the wheel on the left, and as this is rolling over, also the locking arm comes back forward and locks the wheel on the left in place.
So far for the tens’ carry then. But why was I having all the trouble in the world getting the machine to enter numbers? So how does the number entering operation come about exactly? Apart from what’s happening in the bottom part of the machine (we’ll get to that in a minute), essentially what happens is that a kind of spokewheel with long thin teeth pokes up through the top plate of the machine, and engages with the slots in the numeral wheels, pushing them forward.
And there turned out to be the rub. Remember I mentioned that the carriage wouldn’t sit on the machine correctly ? The reason for that is that the little stubby thing sticking out of the top plate on the back of the carriage is not just a guide or a support - it is supposed to slide over the little tails to the locking lever, and push down those and only those that are currently in the active position. That is also why it is shaped like it is, with a flat top, spiky ends left and right, and a curve going down the underside, so that it registers exactly with the locking lever of the active wheel and pushes down on it.
In this way the wheels are always locked, unless they are in the active position, or they are about to undergo a tens’ carry. However, this was apparently not sufficient to exactly register the number in the window of the carriage every time, and at the front of the carriage, for every numeral wheel, an extra strip of spring steel is added that falls in the slots between the numbers, and the wheel gets pushed back exactly into the window by the locking lever because the edge of a number snugs up to the edge of this precisely shaped spring.
So then - why didn’t it want to work at all ? Because what happened was that with the carriage reassembled, when I would try to clear the register, several wheels would overshoot their 0 position.
When keying in numbers, some positions would also not register correctly, and then the movement of the carriage to the left, which happens through the transport mechanism at the end of registering a digit, would also block, because the teeth of the spokewheel would not have a slot between two numbers to slide through, and the entire machine would be blocked. All 6s ?
All 7s ?
After a while, I figured out this had to do with the locking levers. Of course, when looking at the bottom of the carriage when there are numbers in it, the visual appearance of the locking levers is very uneven, because they will be pushed down/back by the carry arms depending on what number the associated wheel is registering. You can also push them all down with the thumb button. But after looking at them for a good long while, the penny started to drop.
When the machine arrived in the workshop, the carriage didn’t sit right onto the machine. In fact, it is not so easy to get the carriage on right - because the “unlocking support” at the back of the carriage needs to be fed through a hole on the right of the carriage, so that it slides its sharp point over the tails to the locking levers - not under. If someone takes the carriage off, and then put its back incorrectly, which is very easy to do, then in the course of faffing about to get the machine to work correctly (which it won’t), the thin tips of the tails get bent. And once they are bent upwards, they will retract the locking lever too far, preventing it from registering correctly when clearing, and the numeral wheels overshoot. So I had to carefully bend them all to line up, and suddenly the clearing problems were solved, and the carriage clears reliably. However, it would still not solve the issue with the higher numbers not registering correctly when a 7, 8 or 9 key was pressed. To understand why this was the case, we first need to look into the bottom part of the machine - as I should have done, before removing the carriage another three or four times because I was convinced it was a problem with the locking levers.
But no! Here’s the rub... The machine works by winding a spring on the right side with a pull lever.
About 17 pulls winds the spring completely. The amount of spring relaxation scales to the total of the digits - it is about 55 units per stoke of the lever. So keying in 9’s across the board will relax the spring about 1.5 strokes. When operating the machine, it is recommended to give it a few strokes after every few numbers entered, to avoid trouble with the spring force becoming too low and the machine too sluggish. The rest of the mechanism is almost impossible to photograph, but what happens is that on every key push, a single toothed block is pushed up in to the teeth of the large central gear, and the spring is released.
This turns the central gear, until the block hits a hook at the back of the machine, connected to the pivot of the key in question (this can be somewhat more clearly seen in the pictures all the way up on top of the page, of the machine prior to cleaning). The earlier along the rotation of the large gear it encounters this hook, the lower the number registered. There are 9 hooks at 9 different distances, each connected to a different key - and that is how the machine counts 1-9. A the end of the stroke, the release of the key operates the escapement, which allows the spring-loaded carriage to slide one place to the left. The 0 key only operates the escapement.
And so when I finally looked down into the back of the machine when keying in some high numbers - it turns out the 8 and the 9 are the furthest away from the central wheel - the two large spiral springs you see, which are supposed to pull back the single-tooth block after it has been transported backwards and upwards by the teeth of the gear, simply interfered with the travel of the arms with the hooks.
Reorganizing them and clamping them between the bar and the frame plates, as they must have been before an idiot got his hands on the machine and screwed everything up, fixed the problem, and the machine now operates faultlessly, even with the spring wound down to half its strength.
And still, while looking very pretty, it remains a finicky, temperamental little thing. I wouldn’t trust it when doing calculations, and when a machine gives you this feeling, you know as a customer as well as a manufacturer that it will never be a commercial success. And that is indeed how it turned out, no matter how nice the mahogany box it sits in is...
