An explanation for an exception to Grimm's Law, which states that as a whole, voiced stops (litterae Mediae, b,d & g) regularly shift to voiceless stops (litterae tenues, p,t & k), which regularly shift to aspirates/spirants (litterae aspiratae, φ, θ, and χ). The Danish scholar Karl Verner examined a special case, in which voiceless stops shift instead to voiced stops instead of spirants, such as the correspondence of a Latin pater to Gothic fadar where we might expect faþar.
Verner, building on the work of Carl Lottner (who had in 1862 already compiled a list of these exceptions), and examining what he called strong verbs, determined that the cause of the anomaly was 'the variable IE accent'. He concluded that f, þ, and χ, when in medial position, derived from an earlier p, t, & k, and not immediately preceded by an accent, become the voiced spirants bh, dh, and gh. So, the reconstructed *upéri gave rise to the Old High German ubar, *sep(t)m to the Gothic sibun.
One of the failings of Grimm's law was that he stated a general rule without examining the exceptions; Verner, along with Grassmann, went a long way in proving the validity of Grimm's work by regulating and clarifying these exceptions, providing regular laws by which exceptions occurred. The results of Verner's study were first published in the article, Eine Ausnahme der ersten Lautverschreibung, in 1876.