Double trouble, oil and dribble (the XPAG rear seal)

Eric Worpe delivered this presentation to the MGCC T Register’s ‘Rebuild’ Seminar at the British Motor Museum, Gaydon, Warwickshire on 10th April 2022. He has kindly let me have the slides and photos, starting with a photo of his rather inquisitive cat (Zazzy), who took an interest in the drawings whilst Eric was preparing the presentation.

The prolific inventor, Charles Kettering, once said “Every great improvement has come after repeated failures, which are the finger posts to success”.

These words form the only encouragement as we try to tackle the leaks from XPAG engines. As over 60% of T-types have survived, the oil leaks must have played a significant role in preserving the chassis and engine compartment, due to the unrelenting spray of oil everywhere. However, as T-types are now increasingly cherished and occupy garages, oil leaks have become an embarrassment, especially on fancy block paved driveways and when the car is being presented for its MOT test. Did you know that an oil puddle greater that 75mm after 5 minutes of engine running is considered a failure?

Any engine sealing system is going to be challenged by the build up of pressure within the engine block. As the engine cylinders and piston rings wear, increasing levels of “blow-by” past the rings occur. Such fumes are vented out of the engine block by the “breathing system”.

In modern engines (post 1960s), crankcase fumes are positively extracted by the vacuum in the inlet manifold, or in some cases, drawn into the exhaust system. Both systems have the advantage of working, even when the car is stationary. These partially enclosed systems were introduced during WW2 to allow vehicles to be driven in deep water.

Ed’s note: Charles Kettering invented the ignition system used in our cars back in 1908. An article on Kettering’s invention and its subsequent development was featured in Issue 57 of TTT 2.

2. The XPAG Crankcase Ventilation System

Drawing of side view of engine showing the breather tube.

The XPAG crankcase ventilation system was based on fumes being extracted, by the venturi effect, from the end of the tappet cover’s breathing pipe by the air flow past the end of the pipe when the vehicle was in motion. Air was probably intended to be drawn in through the rocker box by way of the air filter, thus enabling a dilution of “blow-by” fumes within the crankcase. However, as engines wore, fumes were also forced out of the rocker box, particularly when stationary. due to the absence of any venturi effect.

This build-up of crankcase pressure when stationary can overload the oil seals due to its pulsating nature.

Restrictions to escaping fumes may be due to congealed oil blocking the breather pipe and/or the tappet cover gasket covering up the breather hole in the cover. There should be some raised weld nodules preventing the gasket blocking the breather. Paul Ireland has made up some steel plates to prevent this happening.

Ed’s note: These plates (steel gaskets) and nitrile bonded cork gaskets are still available – please see the item in ‘Bits and Pieces’.

3. The Archimedes Scroll Set-up

The traditional methods of sealing the crankshaft within the crankcase were based on 1, graphited rope seals, which are effective at low peripheral speeds in relatively undemanding locations and 2, The Archimedes screw, a revolving obstacle course that needed careful setting up.

Initially, the Archimedes screw worked reasonably well, thanks to a minimal gap of 2 to 3 thou, set up between the crankshaft’s screw thread and the engine block’s housing.

Drawing of the Archimedes screw set-up.

However, as the engine began to wear, the tight tolerances would open up, leading to more oil being flung from the bearings and more significantly, the potential wearing away of the housing surrounding the screw thread as the crankshaft deviated from running true. Subsequent regrinds of the crankshaft could result in some misalignment between the main journals and the screw thread. Any ‘line-boring’ operations could also create misalignment. Rectifying such misalignments was not attempted in many cases due to its time- consuming fiddly nature.

4. The Mazak Housing (pictured below).

The Mazak cap, surrounding the upper half of the Archimedes screw, was located by two 4mm dowel pins and secured by three M6 screws.

The dowel pins were sometimes removed to help realignment during an engine rebuild, but usually the Mazak cap was just replaced without checking its alignment with the screw thread. The main problem with checking the alignment was due to the visual obstruction caused by the crank’s flange supporting the flywheel.

5. Use of a Dummy Mandrel

This can be overcome by machining a dummy mandrel that’s clamped in the rear main bearing housing and presents an extended shaft to simulate the Archimedes screw diameter within the two housing sections.

Photo of dummy mandrel in position.

The use of the dummy mandrel enables the measurement and possible correction of any misalignment.

6. Adjusting the Mazak Housing.

Assuming the rear main bearing housing sets up a reasonable clearance, the Mazak cap can be adjusted to give a suitable clearance by removing material from either the ends or round the rim (as in the picture above).

As mentioned previously, the dowel pins may need to be removed and the Mazak cap secured by some semi-hardening gasket sealant and the three M6 screws.

7. Excessive clearance at rear main bearing housing.

If the rear main bearing housing shows excessive clearance, then an idea by the late Ray Sales could be used. He wrapped a single layer of Sellotape around the Archimedes scroll, then roughed up and cleaned the surface of the rear main bearing’s housing covering the scroll, before applying a layer of JB-Weld or Araldite. The main bearings, crankshaft and bearing caps were assembled and torqued down. Once the JB-Weld had hardened and set, the assembly was dismantled and any excess trimmed off. On removal of the Sellotape around the Archimedes screw, the clearance around the scroll should be accurately established at around 2 thou – the thickness of the Sellotape.

The original Archimedes screw can be effective when set up correctly by paying attention to detail. Using a setting gauge type mandrel to simulate the crankshaft’s oil scroll makes a significant contribution, unless the crankshaft’s main journals have been ground off-centre. This can be checked using a dial gauge to measure any ‘run-out’ of the crank’s flange referenced to the crankshaft’s main bearing journals.

Measuring the ‘run-out’ of the crank’s flange.

8. Conversion to a modern oil sealing system.

Much has been said about the conversion to an oil sealing system based on the type used in modern cars (in our case, post 1960s). However, modern oil seals used in the XPAG have had mixed results.

My chance to gain some experience came whilst helping a friend with his TC (I’m a firm believer in trying new things out on other peoples’ cars). This experience has resulted in some insight, but also considerable frustration.

Available kits substitute the Mazak cap that sits above the rear main bearing housing, with a semi-circular aluminium housing. A second semi-circular housing is then secured in line with the first, using two tangential bolts and this is then sealed to the face of the main bearing housing on assembly.

The oil seal housing

The complete housing supports a standard imperial oil seal which runs, not on the Archimedes’ screw, but on the crank’s flange supporting the flywheel. The half housing replacing the Mazak cap is secured by three M6 fixings, having removed the dowel pins.

Some kits have detailed instructions on how to ensure concentricity between the oil seal and the flange, other kits assume the positioning of the three M6 screws will set up the alignment.

9. Fitting a replacement oil seal.

This particular engine already had an oil seal conversion which was starting to leak, possibly due to the ‘blow-by’ fumes caused by engine wear; we also needed to access any damage caused by water entering one carburettor after the car had been driven through a deep puddle. The resultant steam generated had caused a pressure pulse that blew the head gasket between the 3rd and 4th cylinders and did an amazing job of cleaning one combustion chamber of all its carbon deposits…..but also bent the conrod.

The replacement oil seal, we were assured, had been “carefully selected” and we were advised to fit a 1mm spacer behind the oil seal in its housing, a clue that some inherent problems existed.

Once the engine was re-assembled and warmed up, we found the oil leak was even greater, so out came the engine again. A closer look at both oil seals revealed the new one lacked a garter spring. This surprised us as the rubber part of the old seal had stiffened from exposure and consequently would have benefitted from the garter spring helping to apply pressure to the wiping ring of the seal (see next drawing “Two oil seal profiles”).

Two Oil Seal Profiles.

However, another more significant aspect concerned the profile of the rubber part of the seal. The inclusion of a garter spring in the old oil seal meant the wiper’s rim was enclosed well within the profile of the seal. Excluding the garter spring in the new seal, resulted in the wiper’s rim being displaced to the edge of the seal’s profile and this forced the wiper’s rim to run on the edge of the crank’s flywheel flange. The 1mm spacer had not been able to correct sufficiently for this. Additional spacers would have moved the oil seal further out of the housing and into contact with the flywheel.

We decided to search for an imperial seal with a garter spring, but only found a NITRILE rubber version whose maximum peripheral speed was restricted to 14M/sec., roughly equivalent to 3,000 rpm. However, as the owner of the engine admitted, he rarely goes below 3,000 rpm, this seal had to be rejected in favour of a VITON seal capable of 40M/sec. Unable to find a suitable imperial VITON seal, we eventually realized, thanks to an article in the Y Register, that an almost equivalent metric seal in VITON is available at 95mm x 120mm x 12mm.

Relative dimensions of seals

The crank’s flange is actually 95mm dia., so ideal for the metric seal. However, the seal’s housing is some 25 thou. greater than the outside dia. of the seal. Our main concern was the width of the metric seal at 12mm, some 100 thou. wider than the 0.375” of the imperial seal. This meant a machining operation to remove 80 thou. from the flywheel face adjacent to the seal.

Machining the flywheel and fitting a “speedisleeve”.

Machining the flywheel

Unfortunately, such an operation would also reduce the depth of the counter bored section of the flywheel that helps locate the crank’s flange; not a good situation.

We also decided a fit a “speedisleeve”, despite the crank’s flange being smooth, as we hoped it would provide a small extension over the chamfer of the crank’s flange.

The “speedisleeve and fitting cup”

As the lip of the seal faces forward, there was concern that sliding the seal over the edge of the “speedisleeve” might damage the seal’s lip. This was resolved by using a tube type guide made from thin plastic sheet or part of a large Coke bottle to cover the edge of the “speedisleeve” and lightly lubricating with Vaseline.

We initially thought the outer diameter of the seal, some 25 thou. smaller than the housing, could be secured by heavy-duty sealant. However, given that the original seal was an interference fit within the housing, we decided to introduce a strip of abrasive paper super-glued to the housing bore. This presented an interference fit of a few thou. to the seal’s outer dia., thus securing a robust and hopefully aligned fixture.

However, only the open end of the seal was enclosed by the housing, leaving the sturdy closed end some 4mm proud of the housing. The interference pressure from clamping the two halves of the housing tended to squeeze the seal out of the housing, any sealant almost acting as a lubricant.

This was overcome by machining up an aluminium retainer ring 4mm thick and fixing this to the housing.

Adding the ‘Ally’ retaining ring

The inner dia. of the ring was shaped to cup the outer edge of the seal.

Armed with a new Viton seal, a 1mm spacer and “speedisleeve”, imagine how we felt about the continuing oil leak when the engine was warmed up!

Where had we gone wrong?

There are conflicting views on the oil seal conversion, so what are the variables that decide its effectiveness? Others have been successful, often mentioning the importance of attention to detail.

Is the Viton seal material unsuitable? We noted the latest offering for the seal is based on a graphite loaded PTFE version which is capable of handling 12,000 rpm. At such a speed in an XPAG, oil leaks would be the least of your problems!

Diagram of assembly problems.

The frustration from the poor outcome encouraged further investigation and one of the first points we realized was that the new oil seal housing fails to replicate the Archimedes scroll provided by the old Mazak housing. The gap was some 2mm or 80 thou. instead of 3 thou. The reason given was the need to allow some oil to lubricate the oil seal, as if that was ever going to be a problem. This really represents the lost opportunity of both oil sealing systems working together.

Looking up the specifications on oil seals, we found that SKF have taken over Chicago Rawhide and offer a 223 page technical brochure in which we learnt that PTFE seals can tolerate dry running but need a hard surface to run on, such as a “speedisleeve”. Given that the Archimedes scroll, even on its best behaviour, still allows some oil through and this combined with the tolerance of PTFE seals to run dry, the retention of the full Archimedes scroll facility, looks promising. This would need only a small change in the design of the seal’s upper housing.

The second area of concern is the location of the bleed hole that allows oil to drain from the cavity formed between the seal’s housing and the crank’s flange. Oil flows through the hole into the trough in the main bearing cap, down the tube and into the sump. This 3/16” dia. hole is pitched just above the seal as opposed to being adjacent to the seal. This means that when static, the lower part of the seal sits in an oil bath.

However, more concerning is the dynamic situation. The crank’s flange sits between 1 and 1.5mm away from the housing, so what effect has the whirring flange, so near the bleed hole entrance, on the ability of the bleed hole to drain away any oil? Could the drain hole be angled to encourage oil to drain away?

We also wondered if the seal could be overloaded by the pressure caused by the whirring flange centrifuging oil into the chamber formed within the seal housing.

Unfortunately, the location of the bleed hole is constrained by the dimensions of the rear main bearing cap, so improving this issue is unlikely. All the more justification for future housings using PTFE seals to reintroduce the Archimedes scroll.

The two oil seals that had had exposure in the engine showed a stiffening of the rubber seal. A worrying aspect, especially if the crank’s flange runs out of true, as a stiff seal would lack the accommodation to keep in contact with a wobbling flange, particularly at speed.


  1. The engine needs to be in good condition to reduce “blow by” past the pistons and excessive escaping oil from the rear main bearing.
  2. Pay attention to detail when using the Archimedes scroll and use a dummy mandrel to set up the correct clearances.
  3. If you fit an oil seal, check the run out of the crank’s flange in case it’s not coaxial with the main journals.
  4. Fit a “speedisleeve” to ensure a smooth and extended surface of the crank’s flange and consider using a spacer to ensure the lip of the seal avoids the chamfer on the flange.
  5. Ensure the housing is central about the crank’s flange.
  6. Consider drilling the drain hole at an angle.
  7. Press suppliers to modify the housing to retain the full Archimedes scroll and use a PTFE seal.
  8. The UK available oil seal kits do not position the oil seal in the best place. Doubt can arguably be cast on the ability of the 3/16” drain hole to drain away the oil held back by the oil seal.

The only effective position for an oil seal would be over the original Archimedes scroll. This actual location is used by those who fit the Chevy 350 oil seal. However, this means “surface welding” over the oil scroll on the crankshaft and then machining the weld to fit a split type Chevy oil seal. This is then located within the original housing, both parts of which have to be machined to support the Chevy oil seal. Reports from the USA seem positive, but do depend on finding a competent welder/machinist.

I cannot say I’m happy with welding the crankshaft, and would consider two semi-circular sections placed over the scroll and their ends welded together. This would cause the ring to shrink on cooling and clamp on the scroll.

Eric Worpe

2 thoughts on “Double trouble, oil and dribble (the XPAG rear seal)

  1. Geoff Piddington says:

    Some 20 years ago, I purchase a TF fitted with a rear crank seal. Happilly, I never had the opportunity to closely examine this modification. An additional modification to the crankcase oil breather system was later discovered which I suspect played an important role in the almost total elimination of oil loss from this engine.

    The modified crankcase breather was copied from what I believe is the oil breather system used by late (UK spec) Morris and MGB 1800 and on the modified TF system, routed the crankcase air flow from the air filter through the standard TF connecting hose fitted with a internal 11/16″ plug restrictor. The original TF breather outlet was removed and an elbow fashioned connecting with both carburettors via hose. In effect, this creates a vacuum controlled crankase oil breather system. A more detailed understanding of the design can be found by examination of the later Morris or MGB 1800 crankcase ventilation systems.

  2. David Smith says:

    I ecenty had to change the clutch plate on my TF. The person who did the oilseal coversion had not applied sealant to the four bolts securing the flywheel. With an oil seal conversion in place the flywheelmounting flange is exposed to crankcase oil pressure. Oil had seeped past the bolts, soaked the gearbox first motion shaft and centrifuged up the inside face of the clutchplate.

Comments are closed.