Question submitted by Janice Rosang
Actually it doesn't! What you're probably noticing is a "rounding" error in the publication of the sunrise time. However the sunrise time does remain virtually unchanged for a matter of about a week or so.
And to explain what is going on I'm going to include a discussion about equinoxes as well - simply because you are the second person to ask this type of question. Basically astronomical equinoxes and solstices' are hard and fast explicit moments in time, but they do not mark exact lengths of days etc at our latitude. Here's why:
The seasonal significance of the winter solstice is in the reversal of the gradually lengthening nights and shortening days. How cultures define this is varied, since it is sometimes said to astronomically mark either the beginning or middle of a hemisphere's winter. "Winter" itself is a subjective term, so there is no scientifically established beginning or middle of winter but the winter solstice itself is clearly calculated to within a second. Though the winter solstice lasts an instant, the term is also colloquially - however quite inaccurately - used to refer to the full 24-hour period.
As humans we generally measure the "day" as the time between sunrise and sunset. Sunrise is defined as when the TOP EDGE of the sun breaks the horizon and sunset is defined as when the TOP EDGE disappears below the horizon. Equinox is measured based on the CENTER of the sun's disk, not at the edge. This difference means that at the equinox moment, the day is actually slightly LONGER than 12 hours.
Not only that, but the earth's atmosphere refracts (bends) sunlight coming over the horizon so that the sun appears to have risen even before it really has. This adds 7-8 more minutes to the length of the day at the equinoxes.
Both of these effects combine so that the true 12-hour day for high northern latitudes can occur several days before the Vernal Equinox and several days after the Autumnal equinox. The effect is lessened for latitudes closer to the equator. And the opposite holds true for those latitudes in the southern hemisphere.
In addition, since the earth is not rotating at the actual moment of the astronomical equinox (can't rotate if there is no time change) then only certain points on earth even see a sunset or sunrise at the ACTUAL moment of the equinox. And then the other end of the day for those locations would be before or after the equinox anyway. So no one location on earth can possibly see a sunrise and sunset at the equinox moment.
Still, this third "effect" does not contribute to the lengthening the day as much as the other two.
Hope this answers it!