I have to chime in after reading Marvin's post. He seems to have put a lot of thought into his theory about EJ-22 head bolts. However, let me share some items I have learned about fasteners while working at General Motors Engineering for 30 years.

Fasteners (nuts and bolts) have been around for centuries. Through trial and error evolved to a reasonable level of technological understanding for the time. What puzzeled designers during the industrial revolution was why some bolts failed in applications and others didn't. Failures for unpredictable reasons. The designers tended to compensate for this unpredictable failure possibility by adding larger and/or extra bolts. This sometimes worked. Not until the 1970's did Fastener technology mature through strain gauge advancments. Additionaly, metalurigical and production technology advanced to a level where a bolts properties could consistantly fall in to predetermined ranges.

Modern Fastener Technology is far superior to the "old School" understandings that persists still. All bolts are springs. Always were. Rotational torque alone (static torque) cannot consistantly place a bolt into design clamp position. The only sure way to place a bolt into proper installation is to measure bolt stretch. This is done on bridges and other structural bolts that use nuts and have both ends of the bolt exposed after installation. A given bolt will be X amount longer after proper installation. Bolt length is measured before and after installation. Typically using a go/no-go gauge. Marvin refered to this normal (and desired) elongation as yeild. Yield is when a material it stretched beyond it's elastic limits. A bolt tightened to yeild is more commonally called a "broken bolt" Modern grade 8 bolts have a tensil strength of around 125,000 psi. That is their Yield point. A bolt must be stretched to provide a clamping load. But it must not be allowed to exceed it's elastic limits either. A bolt needs to clamp parts together while moving through cyclic loading, viberation, thermal expansion and contraction and any other stress present in the application and yet remain within it's elastic limits.

What was not fully understood in decades past was that bolts need to move. A bolt has to stretch during it's operational life in order to maintain it's clamp load on the parts being fastened together. Short bolts have less ability to stretch and stay within their elastic limits that do longer bolts. GM had to place spacers on the steering gear boxes of some of their trucks so a longer bolt could be used. Crash tests showed that the short bolts would break during the crash but the longer ones didn't.

Bolts that are installed into a blind hole (head bolts) do not allow the installer to measure the amount of stretch directly. But the correct amount of stretch can be applied by Dynamic torquing. This is the process called for in the Subaru manual for torquing head bolts. Bring each bolt to a low static torque (using a mechanic's torque wrench) in a predetermined pattern. Then bring each bolt to a slightly higher torque in a pattern. Then turn each bolt 90 degree in the same pattern. Then turn each bolt another 90 degrees. This gets the bolts to the proper installed length +/- 2%

As I recall, my manual says to oil each head bolt before installation. This is also very important. Oil (wax, loctite, soap) lubricates the bolt to reduce rotational friction during static torquing. Without oil the torque wrench will reach torque value early before the bolt has been stretched enough. The affect oil has on torque is dramatic. Take a 1/4-20 bolt and clamp it in a vice with the threaded end up. Place a single washer on the bolt and torque a nut to about 15 foot pounds. Most bolts will take this much. Now loosen the nut some and place a drop of motor oil on the threads under the nut. Try and torque it back to the same 15 foot pounds. It will break before you reach it.

Life would be easier if one could determine the applied torque value on a bolt by loosening it with a torque wrench. Not even close. Additionally, in the application of head bolts, the last bolts loosened will look to be torqued at higher values due to the increasing load on each remaining bolt as others are relieved of their clamping duties.

Reusing head bolts? Absoultly! As long as the bolts are not damaged in any way. If it were a high time engine or a head bolt had broken then by all means get all new ones. By not damaged that means no noticable wear marks on the shank, head or threads. No dings, nicks or scratches. No heat affected areas, deposits or corrosion (leaking head gasket?). Feel free to mix some new bolts with undamaged used bolts if some are suspect. Manuals almost always tell you to install new fasteners that are considered critical fasteners. This is their only way to assure original design compliance for legal reasons. If you are uncomfortable reusing your head bolts or there is any doubt then new head bolts might ease your worries. If they are good then reuse them.

It is important to follow the manual proceedure for torquing head bolts weather you use new bolts or your original ones. Your bolts have not be damaged if they were once, twice or more times installed correctly. If you are worried use a micrometer and look for necking of the bolt. A smooth reduction in diameter mid point or near stress risers. That might suggest your bolts have had excursions beyond their elastic limits.

I looked into this several months ago and concluded that used bolts would work just fine as long as one did not use the torquing method for new bolts. They should be torqued back to the value as originally installed. Also would be best to put the bolts back in their original locations. My reasoning: The final 180° (90 + 90) of torque for new bolts is what yields (permanently deforms) them.

Not true. Yield=failure. You cannot torque a bolt to yield. Once a bolt reaches yield (exceedes it's elastic limits) it continues to stretch until it breaks in two. With dimishing torque resistance.

At this point they become constant pressure springs that provide unchanging clamping pressure (look up Young's modulus)

A spring yes, but not a fixed resistance. Spring tension increases with motion like any spring does.

as the engine heats and cools, in spite of the differences in thermal coefficient of expansion between steel and aluminum. Since the bolts were deformed on the first installation, following the book torquing instructions again would further deform the bolts. "Torque-to-yield" could become torque-to-failure.

Torque to Yield IS torque to failure. Everytime!....

I measured, as best I could, the original torque of the bolts on my early EJ22, which I think is factory original. My conclusion was that the center used bolts should incrementally end up at 85 lb.-ft. and the end used bolts should end up at 65 lb.-ft. , following the torquing sequence. I have not tested this.

I would guess you measured the break away torque of the outside bolts first and ended at the center bolts. This would predictably give you increasing break away values as you proceeded.

I hope this offers everyone some interesting insight on modern fastener design.