Darroll's laid out the math pretty well. You need to look at the intervals. Suppose you have a STAR ewe who is settling and lambing at every opportunity. In year one, she lambs January 1 and August 15. presto, two lambings per year!. The following year she lambs March 1 and October 15. Bingo, she's done it again! The following year she lambs June 1. The next time she lambs is Feb. 15 of the following year. This is based on an average lambing interval of 7.5 months, and I think Cornell's target is actually just a little shorter than that -- 7.2 months or so, which makes the five breeding/lambing times work out evenly within the year so that lambs are born in the same months year after year. My example moves each lambing ahead a couple of weeks every year.
But Cornell will be quick to tell you that it's a very rare ewe who lambs at every opportunity. The average interval is likely to be much closer to 8 or 10 months, so the average ewe in the Star system is really lambing three times in two years. Still quite a feat of management and production, but not twice a year.