On Oct 19th I experienced a partial rappel anchor failure when a Dyneema sling used as part of the anchor broke at the point where it was girth hitched to a wider Spectra sling. The anchor was set on Oct 15th, 30 feet back from the top of the cliff and used only for rappels and was never subject to any sudden impacts. The anchor was equalized for the direction of pull with the girth hitched point of failure on one side. In theory this part of the anchor should have only been subject to half the loads, however because the equalization point was a clove hitch it is possible that one side could have taken the full load (body weight only) if the angle of the rope changed. The rappel line was a static rope and I was using a Grigri as my rappel device. Where the rope ran over the lip of the cliff it was padded with a piece of carpet.Every morning the anchor was inspected for soundness (especially any chewing by rodents) and seemed sound. When loaded, the Dyneema sling was suspended in mid-air with absolutely no contact with the rock to cause abrasion. The anchor was rappelled on a total of seven times (sling failed on 7th rap). As we were cleaning routes, the time of rappels varied from about 10 minutes to one hour. There was a small amount of rain on the 17th and possibly some showers on the 18th, however the rappel line felt dry on the 19th.The failed sling was in use less than one year and showed no signs of damage prior to this incident. The girth hitch feels quite tight after the break. When the sling finally failed I was partway down a slab with multiple points of contact between the rope and the rock to reduce the force on the anchor. I was stationary at the time and suddenly dropped a foot when the sling broke.The point where the Dyneema failed was Dyneema pinching Dyneema and not in contact with the Spectra (see photos). The Dyneema sling that broke was an 8mm Mammut runner. The sling it was girth hitched to was a 5/8? Misty Mountain Spectra runner. The broken ends of the Dyneema feel soft, not fused.What factors might have increased the load beyond body weight?* Pendulums across the face ? 10 foot lateral swing at most, 30 feet below the point the rope ran over the top of face.* Extra force applied when prying loose blocks of with a prybar. In theory this force could not exceed the amount I could deadlift. In this situation I don?t beleive I ever exerted more than 100 pounds of additional downward force on the rope.* Sudden drops onto rappel line after it was unweighted when standing on ledges. At times the rap line was partially unweighted, but because I am nervous/careful when it comes to ropework I always sucked up the rope through the Grigri before reweighting it.* Extra weight of bolting gear, prybar, etc ? 40 pounds at most. I don?t believe any of the above factors is significant, especially since all of these take place below the lip where the friction of the rope on the lip would reduce the load on the anchor.At times the sling may have been stored in the same pack as a Bosch battery. Is there any evidence that NiCad cells emit anything that can damage Dyneema?At present my best guess as to why the sling failed is that when girth hitched tightly such a small diameter sling can cut through itself (the sling suffered a very clean break). In this incident I can imagine that the girth hitch received numerous small tugs under low load (body weight and less) and this might have caused a repeated microscopic nipping or sawing action that eventually cut through the whole sling (the wider Spectra sling it was girth hitched to also had some fibers cut at the point of contact with the Dyneema). I am trying to contact the equipment manager from Mammut to have this sling failure expertly analyzed. Until then I suggest that Dyneema slings should never be girth hitched or otherwise knotted.John Sherman