Some of the variation can be explained by the inclusion or omission of Sunday's sales - only Friday and Saturday fell into the quarter ending June 30. But even allowing for that, how do you reconcile Merrill Lynch's "at least a couple hundred thousand units" with Goldman Sachs' "at least double our prior 350,000 estimate"?

Imagine Friday evening's sales were, despite the hype, only half that of the full-day trading on Saturday and Sunday. Then June quarter sales of 200,000 become something like 350,000 for the weekend. Add a 50 percent fudge factor for "at least", and the resulting 525,000 is still way short of Goldman Sach's estimate. On the other hand, it is very close to Piper Jaffray's 500,000, which we know was based on a survey, and exactly the same as an estimate published by Global Equities Research.

Similarly, variations in the longer term estimates could be due to the use of either fiscal or calendar years.

Unless iPhone sales hit the 'magic million' this week, Apple probably won't disclose any numbers prior to announcing its  quarterly results towards the end of this month. Even then, we would not be surprised if the company lumps the iPhone in with the iPod figures unless the opening Friday/Saturday sales really were spectacular.