The tensioner is installed in the right front corner of the Jaguar V12 engine, applying tension to the timing chain by pressing against the slack side of the chain.  It is held in place by a pin through the large hole at the right end in this picture, and the "shoe" on the bottom of the left end rests on a surface within the cover and can slide on it.  The chain rides over the arched upper surface.  The pinned end is down near the crank, while the shoe end is up near the right side cam sprocket.

The latch is the device in the middle with the slotted holes in it.  Here's a closeup of the latch itself, and here's another closeup of the latch.  The spring pulls the shoe towards the pinned end, causing the arch to bow and press farther into the chain line, and the rod slides through the latch.  The rod cannot back up, however, because the latch rocks (clockwise in these pictures) and the edges of the hole dig into the sides of the rod and grab it.  Note that this non-return feature may appear to be of little importance since the spring is maintaining tension anyway, but the latch is there in case operation causes dynamic loading or the engine is turned backwards for some reason.  A little chain whip pushing the tensioner back and allowing the chain to "jump time" would not be good.

Manually rocking the latch back (ccw) with special tool JD.50 releases the rod and allows the slider to be pulled back to flatten the arch and release tension on the chain.  To lock it in the retracted position, the slider is pulled back until the smaller diameter section at the end of the rod pulls into the latch.  The smaller diameter rod in the hole in the latch allows the latch to rock farther.  By rocking the latch farther ccw, the latch grabs the step in the diameter of the rod and holds the retracted position against the spring tension.

A few notes about this particular part.  It came out of Kirby Palm's '83 XJ-S in the late 80's and has been sitting around ever since, waiting for access to a digital camera -- that's why there's some "patina" on the metal parts.  Yes, it's broken; the arch itself was three separate pieces until it was superglued back together for taking pictures -- you can see the cracks.  The spring has been disconnected, since it will happily snap the arch back to pieces; it is normally supposed to be connected to the pin that is visible partially inserted at the right end of the arch.  The brown plastic sleeve is supposed to be installed around the spring.

This part had about 60,000 miles on it at removal.  There are three wear grooves visible along the top surface, corresponding to the end plates on the double-wide chain.  The wear was not a significant problem here; this part broke while Palm was trying to release it after some head work.  In fact, there are no reports of wear being an issue with this part; its problems are always due to fracture, not wear.

If you look carefully, you'll note that the latch rod is slightly bent -- apparently a common malady.  It's supposed to be straight, and it's not too likely that it gets bent during operation; more likely, a mechanic trying too hard to rock the latch bent it.

Here's another view from a different angle.

Here's a closeup of the shoe end, and here's a closeup of the pin end.  The larger hole at the left end of the shoe is where the tensioner retractor tool is hooked to pull the tensioner into the retracted position.  Note that the tool's hook cannot have too much tang on it, or it will interfere with the end of the arch itself and the pin that connects it to the shoe; that's why the book shows how to trim the hook so it will fit properly.

Even though this part receives more than its share of epithets from mechanics, it is actually an excellent design.  The problem with this part is the material the arch itself is made of -- which should come as no surprise, since it is a British nonmetallic part.  The plastic arch gets brittle with age and/or heat, and then shatters at the slightest provocation (or when some innocent mechanic tries to retract it) and causes a considerable increase in the mechanic's workload.

Did I say British?  Maybe not.  Martyn Sandbrook points out that what appears to be the same tensioner was used on the Saab H type 8-valve engine, and provided this picture from a Haynes manual:

OK, the diagonal struts are straight in this picture and elbowed in the Jag part pictured above, but the arch itself sure looks similar.  The idea is obviously the same.  So, perhaps both tensioners are made by the same vendor somewhere.  Perhaps the arch from the Saab tensioner could be used to rebuild the Jaguar tensioner.

Sandbrook also reports that the same tensioner was used in the Jaguar 2.9 litre two-valve six cylinder, although it was not used in the four-valve AJ6 engines.

Since Jaguar produced the V12 for more than two decades without ever seeing fit to correct the problem, what is really needed is an aftermarket replacement for this part.  There's no shortage of ideas for improving it; perhaps the simplest method would be to replace the arch itself with a part made of spring steel.  Spring steel is used as a tensioner in other engines such as the Triumph 6-cylinder and apparently works well enough, although it presumably would make more noise than the plastic part.  However, providing the five attachment points on a piece of spring steel is not the kind of thing a home mechanic can finagle, so even this simplest of upgrades will require a manufacturer that knows how to make such parts.

An improved version of that idea would be to provide a facing on the spring steel, such as a layer of nylon or Teflon or something.  This doesn't make it any easier to make the spring steel part, and figuring out how to affix the lining just adds to the complexity of producing the parts.

A more revolutionary idea would be to replace the entire tensioner assembly with a rigid arch that pivots around the mounting pin.  To apply tension to the chain, a spring would rotate the arch up into the chain path, and some sort of latch would need to prevent it from backing up.  With a little care, it might even be possible to design the latch/spring assembly so that it can be retracted via the 1" diameter hole in the timing cover with no need for removing the right side cam cover.

A variation on that last idea might be to use the stock arch in such a design.  A stiffener could be fabricated that fits against the bottom of the plastic arch and is perhaps attached to it at the original attachment points for the latch rod, spring, etc.  With the arch held rigidly so it cannot flex, it should last indefinitely.

There is one possibility for a minor alteration that may provide some additional durability in this part: grind off the end of a pin and disconnect the arch from the end of the shoe (left end in the pictures above), and allow the bottom corner of the arch to merely slide on the inside surface of the shoe even as the shoe slides on the timing cover.  Actually, it is probable that doing such a mod properly would require making a new shoe a little longer than the original to prevent the end of the arch from dropping off the end of the shoe.  Once disconnected in this manner, the original spring and latch assembly should continue to work as intended, with the elbow-shaped links from the latch end of the shoe to the midspan of the arch continuing to force the center section of the arch into the chain path and taking up the chain slack.  The arch will still need to flex, obviously; the effective difference is that the shoe can no longer apply a tension or compression stress on the arch, so the only stress on it is bending.  Will this actually help prevent it from cracking?  Well, it will probably break eventually anyway, since it seems to be its nature to get more and more brittle with age, but perhaps it will last longer before it cracks.  That is, unless unpinning that end results in some additional stress for some other reason.

Lots of people seem to feel that the best way to design a tensioner would involve a small sprocket on a spring-loaded lever.  However, this would need to be carefully done to avoid dynamic chain line problems on a 6500 RPM engine.  All of the other straight sections of the chain line in this engine are fitted with damper plates to prevent whipping, but the arched tensioner itself is intended to prevent whipping on its section.  If a small sprocket is used instead, there could be chain whip between this sprocket and either the cam or crank sprockets.  It may be possible to provide damper plates alongside a sprocket, but the plates would probably need to move with the sprocket in order to maintain an effective clearance from the chain.  Add the fact that bending the chain around a small sprocket would wear the chain more than the arch does, and you start to lean back towards an arch design.