Scaling of voids and fractality in the galaxy distribution

We study here, from first principles, what properties of voids are to be expected in a fractal point distribution and how the void distribution is related to its morphology. We show this relation in various examples and apply our results to the distribution of galaxies. If the distribution of galaxies forms a fractal set, then this property results in a number of scaling laws to be fulfilled by voids. Consider a fractal set of dimension D and its set of voids. If voids are ordered according to decreasing sizes (largest void has rank R = 1, second largest R = 2 and so on), then a relation between size Lambda and rank of the form Lambda (R) proportional to R-z must hold, with z = d/D, and where d is the Euclidean dimension of the space where the fractal is embedded. The physical restriction D < d means that z > 1 in a fractal set. The average size (Lambda) over bar of voids depends on the upper (Lambda(u)) and the lower (Lambda(l)) cut-off as (Lambda) over bar proportional to Lambda(u)(1-D/d)Lambda(1)(D/d). Current analyses of void sizes in the galaxy distribution do not show evidence of a fractal distribution, but are insufficient to rule it out. We identify possible shortcomings of current void searching algorithms, such as changes of shape in voids at different scales or merging of voids, and propose modifications useful to test fractality in the galaxy distribution.