Note that a date/time value is actually stored as a Double Float
number, a count of days and fractions of a day since midnight,
December 30, 1899; it can be formatted any which way, sorted
chronologically, etc.
I've found it useful recently to use the term "string representation
of dates," which is what the formats are. The raw numeric
representation is hard to see, unless you use the "General Number"
format to display it in its raw form.
All the other formats that are human-friendly are string
representations of the underlying date values and when doing
calculations, need to be converted back to the underlying numeric
value. Given that there are so many ways to format the string
representation, and then different ways to interpret it (e.g.,
3/2/2010 is either in February or March, depending on the
interpretation), it's important to realize that whenever you are
doing calculations or writing criteria for a SQL WHERE clause, you
need to use either an unambiguous string representation of the date,
or use a precise date value (using DateSerial() or CDate()).