I agree with msucog. The smaller size samples are actually easier to protect and do not get "thrown" around. Also, it's interesting to note that the rationale behind the three 4x8s versus two 6x12s is not really based on any data. There are hunches and a lot of people have data on correlating the two sizes but no one has taken the time to write anything up and design the experiment so that it stands up to statistical standards. There will be a push in the near future to produce that document. It's part of a two step process. First, remove the "when specified" language in ASTM (it's only there because there wasn't a precision and basis statement for 4x8's). Second step, produce a 30 sample multi-region batch-to-batch statistical document showing that the batch-to-batch variables are identical whether using 4x8s or 6x12s. ACI just took a conservative approach for ACI 318-08 when choosing 3 cylinders instead of 2 for 4x8s. When in doubt, be conservative and error on the safe side. Unfortunately, the reduction to two 4x8s in ACI 318 probably will not occur until 2011. Until then, you can always get approval to reduce it to 2 cylinders from the structural engineer if they are willing.
And in regards to volume, we have implemented this and the space savings are tremendous, not to mention the greater ease in handling and processing.