The late infall of dark matter onto a galaxy produces structure (such as caustics) in the distribution of dark matter in the halo. We argue that such structure is likely to occur generically on length scales proportional to l ∼ t0vrot, where t0 is the age of the universe and vrot is the rotation velocity of the galaxy. A set of 32 extended galactic rotation curves is analyzed. For each curve, the radial coordinate is rescaled according to r →r ≡ r(v0/vrot), where we choose v0 = 220 km/s. A linear fit to each rescaled rotation curve is subtracted, and the residuals are binned and averaged. The sample shows significant features nearr = 40 kpc andr = 20 kpc. This is consistent with the predictions of the self-similar caustic ring model of galactic halos.PACS numbers: 95.35.+d, 98.35.GiThe rotation curves of most spiral galaxies are approximately flat, i.e. (v(r) constant) for r much larger than the disk radius [1], where v(r) is the circular velocity of gas at radial coordinate r. The flatness of galactic rotation curves implies that galaxies are surrounded by halos of dark matter, and that the halo density falls off approximately as ρ DM (r) ∝ 1/r 2 . The curves are far from exactly flat, however. They have all sorts of irregularities which may be referred to as "bumps". When a bump occurs at r >> disk radius, it indicates structure in the dark matter distribution, i.e. a deviation from a perfect ρ DM ∝ 1/r 2 law. We are motivated by the possibility that such structures have universality and that a statistical analysis of well measured extended rotation curves may reveal that.In refs. [2,3], 32 extended galactic rotation curves were selected and analyzed to test the validity of a modification of Newtonian dynamics (MOND) as an explanation of the "dark matter problem", i.e. the discrepancy between the flatness of galactic rotation curves and the exponential fall-off of galactic luminous matter distributions. We analyze the same set of rotation curves with a different goal, namely to search for structure within dark matter halos. We adopt the conventional wisdom that a rotation curve is a tracer of the galactic mass distribution according to the laws of Newtonian dynamics. However, the selection criteria used by the authors of refs. [2,3] that each rotation curve is an accurate tracer of the radial force law, and that it extends far beyond the edge of the luminous disk -are appropriate from the point of view of our analysis. We do not know of another data set of comparable quality. We use the complete data set described there, without any cuts of our own.One broad argument why structure may be present in galactic halos is that the halos form as a result of the infall of the dark matter surrounding the galaxy [4]. If the dark matter particles are collisionless, they oscillate back and forth numerous times before they are virialized by inhomogeneities in the galactic mass distribution [5].