The basic issue is that of the overhang mandates - the small non-proportional skew in the German mixed election system. To explain, first let's see the German election system in an imaginary country with minimum 100 parliamentary seats.

There are half as many directly elected candidates as minimum seats: i.e., in this case, 50. But, normally, a party gets as many seats as his share in the list vote.

For example, if a party got 40% of the list vote and its candidates were victorious in 15 districts, they'll get 40 seats: they can add 25 people from the party list to the 15 who were directly elected.

However, it may happen that a party wins a much higher share of direct mandates than list votes. Say, 40% of list votes, but 45 direct mandates. 45 - that's 5 more than the allotted 40! In the German system, no seats will be taken away from this party or from the others, instead, the parliament increases by the difference - in this example, to 105 seats. Those 5 are the overhang mandates.

But what complicates the picture is that list vote shares and thus overhang mandates are counted separately for each state. However, a system is needed to balance the fractionals. To illustrate the problem, if a state would have 10 seats for list votes and one party gets 44% another 56%, distributing those 4:6 shifts the real wheights - and if there is a tendency, these shifts add up and marr the result. So instead, there is a complex system of shifting one or two seat allocations between the states. Say, if we move one away from the above state to another, 4:5 will be much closer to the real result in that state.

And what can happen in Dresden is this: if the CDU gets more than 41,000 list votes (in 2002 they got 50,000), that not only means one less CDU overhang mandate in Saxony (2 instead of 3 now predicted), but one more list seat for Saxony and one less for North Rhine-Westphalia state. Since the list seat lost in NRW would be the CDU's (its currently 47 seats there represents an up-rounding), and the one gained in Saxony would mean no change for any party, the end result is -1 for the CDU.

To summarize, a higher CDU vote would lead to a different balancing of fractionals between the states, which would lead to the loss of one overhang mandate in Saxony.

Now, as the article writes, there are two more (but not really realistic) scenarios that would shift votes. First, and this could happen in combination with the above, if the SPD doubles its list votes from 2002 (to around 45%), it would gain one list mandate from the Greens. And if, against expectations, the SPD wins the direct mandate in Dresden instead of the CDU (which reduces the CDU's projected overhang) and achieves some 90% of list votes, it could level with the CDU.

_{*Lunatic*, n.
One whose delusions are out of fashion.}