It has proved to be useful to start with a population of not randomly created individuals. Here a iterated matching approach is implemented which yields densely packed meta-blocks in the individuals. For this, all possible pairs of modules are built and valued by the wasted space inside the resulting meta-blocks. Then a set of pairs with minimal overall waste is chosen to built the elements for the next iteration of matching. This iteration results in a set of densely packed meta-blocks with four blocks inside. The matching-process is iterated until the slicing-tree for a single individual is completed.