Holtzmann's law is a Proto-Germanic sound law originally noted by Adolf Holtzmann in 1838. It is also known by its traditional German name Verschärfung (literally: "sharpening"). (A similar sound law which has affected modern Faroese, called skerping in Faroese itself, is also known as "Faroese Verschärfung" in English.)
The law involves the gemination, or doubling, of PIE semivowels (glides) * -y- and * -w- in strong prosodic positions into Proto-Germanic * -jj- and * -ww-, which had two outcomes:
The process is brought about by the fact that vowels (or semivowels) in the syllable margin are invariably transformed into consonantal articulations.
The conditions of the sound change were long debated, since there was a seemingly random distribution of affected and unaffected words. At first, dependence on word accent was assumed, parallel to Verner's Law. One currently accepted solution, first proposed by Smith (1941), postulates dependency on the presence of a PIE laryngeal, which when lost, triggered lengthening as if the semivowels were vowels, and forced them into the syllable margin.
According to Lehmann (1955), the lengthening occurs in the contexts of PIE * -VwH-, * -iyH-, * -ayH-, * -aHy- (where V is any short vowel, and H is any laryngeal).
For example, PIE *drewh₂yo → early Proto-Germanic *trewwjaz 'trustworthy, faithful' →:
One instance where a laryngeal was never present is PIE *h₂ōwyóm 'egg', but after the loss of * -w-, the * -y- shifted into the syllable margin, giving:
Alternative views
Some linguists (e.g. Joseph Voyles) hold that Holtzmann's Law represents two separate and independent sound changes, one applying to Gothic and another to Old Norse, rather than being a common innovation. This is supported by James Marchand's observation that a Runic inscription (niuwila on the Naesbjaerg bracteate of the 5th century) and an early loan into Finnic (*kuva 'picture', cf. Gothic skuggwa 'mirror', Old High German skūwo 'look') do not exhibit this change. If true, this would prevent Holtzmann's law being used as an example of early Gotho-Nordic unity, in which context it is often cited. Voyles's explanations of the changes do not involve laryngeal theory.