We show that the inter-cloud Larson scaling relation between mean volume density and size ρ ∝ R−1, which in turn implies that mass M ∝ R2, or that the column density N is constant, is an artefact of the observational methods used. Specifically, setting the column density threshold near or above the peak of the column density probability distribution function N-PDF (N ∼ 1021 cm−2) produces the Larson scaling as long as the N-PDF decreases rapidly at higher column densities. We argue that the physical reasons behind local clouds to have this behaviour are that (1) this peak column density is near the value required to shield CO from photodissociation in the solar neighbourhood, and (2) gas at higher column densities is rare because it is susceptible to gravitational collapse into much smaller structures in specific small regions of the cloud. Similarly, we also use previous results to show that if instead a threshold is set for the volume density, the density will appear to be constant, implying thus that M ∝ R3. Thus, the Larson scaling relation does not provide much information on the structure of molecular clouds, and does not imply either that clouds are in Virial equilibrium, or have a universal structure. We also show that the slope of the M–R curve for a single cloud, which transitions from near-to-flat values for large radii to α = 2 as a limiting case for small radii, depends on the properties of the N-PDF.