1982 IBM Correcting Selectric II

My 1982 IBM taken on my 1982 Nikon F3/T using Kodak Tmax

The renown IBM Selectric II was released in 1971, ten years after the wildly successful Selectric I, which exceeded sales projections by the hundreds in a mere six months.  The IBM was the precursor of personal computer interface, the gold standard of late 20th century productivity, and an icon of the written word.  The first IBM typewriter paved the groundwork for an empire.  It employed a single type element of 64 characters commonly called a "golfball".  An element strikingly similar to the rotating drum of the Bickensderfer typewriter from the late 1800s to the early 1900s.  While the idea of single element typewriters had been toyed with many times before, in regards to the ease of swapping typefaces, IMB was the first to create a truly remarkable single element machine with unparalleled speed and consistency.  It could be related to the Hammond, and the later Varityper, which were both perfected models (Hammond by 1890-1910), but the visibility, reliability, and the versatility of the IBM exceeded both.

The Selectric II is considered by many to be one of the finest operational IMBs, though I myself prefer the styling of the earlier Model 1.  The Selectric II was introduced with two major configurations: standard, and correcting.  It was the latter feature that set aside the II from the I.  This allowed the user to backspace over a typo, and erase it from the page with a single keystroke, thus eliminating the need for white out and erasers.  To top it off, it contained an easy to install carbon ribbon cartridge, a rare find on manual typewriters.  Carbon ribbons give unparalleled print quality, and their only drawbacks are price and security.

To briefly address carbon ribbon security, it must be understood that carbon ribbons consist of a thin cellophane strip with a layer of carbon that gets pressed into the page.  This leaves a negative space on the ribbon in the shape of the typed character.  Since that section of ribbon no longer contains usable carbon, it is advanced one whole space.  One could unravel the entirety of the ribbon and have a readable record of everything it was used to type.  Back during the cold war, a British double agent was uncovered due to his carelessly discarded carbon ribbon.

The Selectric II also employed the interchangeable golf ball system, as well as variable typing pitch of 10 or 12 CPI adjusted by a rotating carriage rack.  While the Selectric features are numerous, the components to make it work are more so.  

Technological advancements have historically merely been a matter of making things smaller, the perfect example being the inception of the CD, by which a groove is etched into a metalic foil by a laser.  It's larger predecessor, the LP was a groove etched by a diamond, and the LP's larger predecessor was shellac etched by a steel needle.  In order to create a more complex piece of technology, the components must be made smaller.  Similarly, the first IBM computer was a monstrosity of mechanical components, and the Selectric was a mere small portion of all that.  It is at its very core, a 6 bit binary to analogue mechanical computer.  

I personally found the whiffletree to be an ingenious device.  It is based off of geometric principles, and uses basic single bit inputs to generate specific systematic movements.  This device originated as a force distributer from agriculture, designed to help distribute the pulling force of working animals.  The IBM whiffletree is much more complex, and a great video on how it works can be found here.

The Selectric Whiffletree, head tilt on the left, and rotation on the right

And with that video shown, there remains very little need for me to further explain how this machine works.  

As far as my experience with Selectrics go, I never thought I'd own one.  They seemed to be one step more complex than what I was willing to work on, as electronics aren't my thing.  Though given how Selectrics are completely mechanical, and how many clients I've turned down, and the potential of creating a short film based off a story I wrote, I decided it was high time to get myself a Selectric and learn the craft.  Of course I'd love to have a red correcting selectric II, but I'd settle for a black one with black keys, which is precisely how I ended up with two.  Two for parts, non-working IBMs.  The first was tan with black keys, and the second was black with light grey keys.  My goal was to fix both, and swap the body shells...maybe.  At the time of writing this I am 50% done with the tan one, and have yet to even see the black one (which shipped yesterday).

My broken pulley, plus tab rope

The tan IBM arrived complete with a bag of parts.  It had a broken tab pulley, which I take is common.  The base consists of old brittle plastic under high spring tension.  It also had the tab cable (which advances the carrier) and some correction tape.  I needed to supply the missing carrier return cable, which is a little longer in length.  The lengths on these two ropes aren't needed to quite the same degree of perfection as the tilt and rotation tapes.

There are two listed types of cables.  One length for a Tab, one for either tab or carrier, and one for carrier only.  It was clear to me by looking at the machine that the carrier return cable would need to be longer, and since my tab cable measured out to 22", I cut my carrier return cable to be 28".  For that I used the inner strands of paracord, which I knotted one either end with large knots (important for later).  I also ordered a new tab pulley from ebay, and a hand cycle crank which is necessary for installing both cables.  While the IBM tech manual shows some nice diagrams, the video made by Duane Jensen is the best available resource by far.  I will go over what I did momentarily.

Coming back to this (3 weeks later) I find out that the cord I used to replace the CR cable was too stretchy, and wouldn't allow the machine to backspace.  Thanks to Duane once again for clearing that up.  I ended up getting a black IBM that was smashed in the mail, so I salvaged the CR cable from that pitiful machine and the backspace worked fine.  Note: There are two adjustment screws for this mechanism to increase travel, but if they are tightened down too far, the backspace won't latch.

Backspace adjust, just above the mainspring 

The first thing I did when I got the machine, was to remove the metal body coverings.  Super easy to do.  The top lid swings up, and there are two small silver latches right above each platen end.  Press those and the platen lifts out.  Then on either side on the inside of the machine body cover, there are two larger leavers that unhook the entire top half of the machine.  Finally, the machine can be released from the lower body shell with the leaver just underneath the bell on the left end of the keyboard.

Once I took care of that, I began to work on cleaning the machine.  These Selectrics gum up quite easily, especially when the foam padding deteriorates.  Make sure to clean the foam out from under the keyboard, that can be done with an air compressor.  When I first powered this machine on, about 2 thirds of the keys did not operate.  That was mainly due to gummed linkages.  There are three or four main contact points where the keys will stick or not operate.  The first is right underneath the front end of the keyboard.  Each key leaver has a hairspring that gives it a little boost back up.  Make sure they are all clean and resting on each of the leavers.  Back up top, just under the margin bar is the main key leaver pivot point.  That needs to be flushed as well.

Underside springs

Finally what is probably more important are a series of interposer latches that are also held down with springs on top by the margin bar.

One of these is not like the other

Each of these blue springs holds down a tab, that tab is what latches onto the interposer and trips the whiffletree.  If it's not under the spring, the key won't do anything when pressed.  Additionally, they need to move freely.  If the tab is gummy, then the key will stick on the downstroke.  So move the spring to the side, and flush that tiny tab out while working it back and forth to help remedy sticking keys.

By the time I had remedied the issues I could tackle without parts (i.e., cleaning) I had yet to receive my IBM cartridge from Staples.  Doing what I generally do, as you might have seen with the Blick 9 (though I type this article before even beginning to type up the Blick), I used a sheet of paper half coated in fine drawing charcoal, which I then fold in half charcoal inward.  It creates a primitive sort of carbon paper that allows me to test non-fabric ribbon machines.

Lied, this test was graphite pencil from the Oliver Pencils I sell here.

As you can see, the IBM indexes letters just fine on the capital setting, but once you switch to lowercase, there seems to be an issue.  From what I gather, there are three causes.  Improper length of the rotation tape is not the culprit, as it indexes fine on the second hemisphere.  The tilt is also fine, as I tested an entire row of letters top down, and even though rotation was off, they all printed on the same line.  No, the culprit was either improper setting of the shift pulley, shift pulley bearing wear (which is common), or as stated before the tape, but that wasn't the cause.  The easiest way to test of course was to place my thumb on the shift pulley and apply a dot of pressure as I typed lowercase.  The letters indexed on their own just fine, so I took the machine body off and adjusted the setting screw.  The other thing that can go wrong is the type head, or gummy tape pulleys.  On my machine, the head was completely stuck on.  What I ended up having to do, was cut off the top plastic cap, pry out the spring, and then type with the ball until it broke itself free.  The teeth, index pin, and the post all looked alright, so I cleaned and lubricated the head, and the took a look at the rotate pulleys.  The pullies on either side were stiffed up, and after some work on cleaning out the bearings, the machine began to index better.  Still has issues, but until I finish the rest of the machine, I will avoid adjusting the tension of the tapes underneath the machine.  

Here you can see the shift pulley which sits on a vertical axis

(center fame)

With the machine power cycled on caps lock, the adjusting bolt is visible

And then the charcoal test, prints fine.  Note the improper

letter spacing is from me hand pushing the carrier, it is a little stiff,

so I will lubricate the tracks better after I install the needed cables.

Installed a carbon ribbon, here are the results from cleaning the

pully systems. People consider these expensive, but at around 10-20$

which is fabric ribbon average, and a yield of 500k characters,

it really isn't that bad.  0.0004 cents a letter around...

The next step was to install the tab and CR cables, an operation which was not easy.  It required two original IBM cables, one which hooks onto the left underside of the carrier and stretches around two pulleys to the back of the machine where it winds around the rear drum, and another which hooks onto the right end of the carrier and feeds around the tab pulley to the tab drum.  The shaft that holds the two drumps is then placed under mainspring tension, which drives the advancement of the carrier like any other typewriter.

The moment I turned the machine on, all hell broke loose.  The tab cord unhooked itself and tangled around the return pinion, dislodging the CR pinion spring and bending it beyond repair.  That rendered the return function of the machine beyond useless, and required pulling the operational camshaft to replace the spring.  Not for the faint of heart, but luckily there are only 6 screws to undo.  It required another week of waiting to get a set of IBM bristol wrenches, and another visit to ye olde parts machine.

Upper right is the grease stain from resting the 

shift pinion spring on the page.

Instructions for such an operation are hard to find, but I did find a page scanned in an old tech manual that gave a few steps to follow.  As to what those parts where, I had to scour the internet and pick the brains of some old registered techs to find out.  I ended up printing off a super low res scan of the OP shaft assembly, and labeled all the concerning parts.  A scan of which you can see above.

Essentially there are six set screws on the OP shaft.  Two in the torque limiter hub, the black plastic bit on the left, two collars on either end of the tab pinion, one in between the cams, and one far right to hold the shaft against the shift bearing.  Those require either a flathead screwdriver or a bristol wrench.

Torque limiter, torque limiter hub, and tension spring.

This assembly comes off the left end of the shaft

the two index pins for the motor pulley slot into the torque limiter. 

That is what allows the OP shaft to spin.

In the image above, yo can see the fine coiled dark grey spring on the (left) CR pinion.  There is a tiny clip called the CR retainer clip that prevents the pinion from sliding to the right, that gets removed.  The pinion spring retainer clip holds one end of the pinion spring tight against the torque limiter hub, and allows the return clutch to tighten the pinion spring around the pinion, and uses friction to drive back the tab drum (big fat yellow one).  

Once the shift E clip, plastic gear bit, and shift pinion spring were off, I used the hand crank to pull the losened shaft to the right.  I honestly felt like I was going to break it as I wrestled the torque limiter off the left end of the shaft, and in turn slid the CR pinion spring off, but once I got it out it was pure satisfaction.

Victory!

The next step was to throw this piece into my working IBM, which was somehow more difficult than getting the part out of the smashed IBM.  It didn't help that I misplaced the pinion spring retainer clip.  I took it off the machine, and set it somewhere within arms reach, and somehow I still can't find it.  That's what parts machines are for I guess.

Reinstallation tip, all OP shaft flat spots for set screws are on the same side.  There is a blank face between the OP shaft support collar and the shift bearing collar for you to line things up.  Remove the screw from the cam ratchet all the way, and locate the flat spot through the screw hole, you will see the chewed metal bit from where the set screw used to sit.  Line it up, and screw it back down.  As you maneuver the shaft back in place, the cams will go in easier.  There are lower latches and cam followers (4 in all) to move to allow it to go back into place.  

Sample test sheets

Disaster struck once more as I failed to seat the torque limiter pinion close enough to the CR pinion, and it jammed, sucking the pinion spring I just replaced into the gap and bending it.  The return now goes along half heartedly, but not bad enough to the point of ordering a new spring and doing it again.  Good enough is sometimes just that.  It also failed to index lines consistently, but that was solved by cleaning out the right hand platen collar so it rotated freely.

You can see the coil that got sucked into the gap

first full page test with everything back together.

Of course one of the final bits to install would be the two internal plastic dust covers.  Just by looking at it one might come to the conclusion that they don't do a whole lot, and really, they don't.  To be honest, it's probably only there to prevent dumb little kids or animals from sticking their fingers down into the moving parts.  I probably won't even bother as this machine is trouble enough as is.

The crisp carbon print looks absolutely amazing, especially on this lovely Strathmore 100% Cotton paper.  All in all I expected to be done in an afternoon, and instead it was three long weeks fueled by anger and multiple $20-30 purchases.  I'll probably edit this in a week or so and add videos below, but it is time to rest and consider this a job well done.  I have also been working on a Royal O model and an Underwood Universal, and have two super super exciting machines coming in.  Those are a surprise ;)

Adam Savage and his IBM Selectric II