In late 1912, Fritz Goos at the Hamburg Physikalisches Staatslaboratorium discovered a systematic dependency of arc-spectra wavelengths on the length of the electric arc used and on its electric parameters, such as, for instance, the current employed. In early 1913, at Heinrich Kayser's better-equipped physical laboratory in Bonn, Goos was able to confirm these effects using a large concave Rowland grating. He was able to establish that variations of between 3 mm and 10 mm in the length of the arc (...) produced wavelength differences of up to 0.02 Å violet shift and -0.007 Å redshift respectively. Further inquiry also revealed a dependency of the wavelength on the region of the arc selected for spectrometric observation. All these surprising effects were soon collectively named ‘pole effect’.As is shown in this paper, the pole effect threatened the validity of the results of the entire research tradition of high-precision spectroscopy which, around 1910, had excelled in establishing several internally coherent systems of wavelength assignments. These wavelength catalogues had been established by spectroscopists such as Heinrich Kayser, Paul Eversheim and their co-workers in Bonn, by August Herman Pfund in Baltimore, and by Charles Fabry and Henri Buisson in Marseille under the aegis of the ‘International Union for Co-Operation in Solar Research’. They had all produced locally consistent, “homogeneous” systems of wavelengths with estimated errors sometimes smaller than 0.001 Å. However, long before 1913, strange non-local inconsistencies had emerged between these systems that were of much greater magnitude than the estimated error. The discovery of the pole effect opened up the possibility that variations in the arc parameters used in the measurements, which the different teams had hitherto not specified, were responsible for the systematic differences, in their respective sets of measurements, coming to up to 0.025 Å.This paper explores the interrelations between local knowledge production, the strategies for the establishment of local coherence, and the ways in which the community of physicists and spectroscopists handled a possible threat to this coherence after 1913. Around 1930, a general agreement was reached about the physical cause of the pole effect, namely Stark effects, caused in turn by intermolecular electric fields of ions in the arc. Much before 1930, however, the community had already succeeded in standardizing the instrumentation used in high-precision spectrometry and had conformed its research practice to such an extent that from 1917 on the pole effect could be routinely circumvented in high-precision spectrometry and interferometry. Thus, experimentation along with its instrumentation, indeed had ‘a life of its own’, independent of the many unsuccessful efforts to explain the pole effect theoretically. (shrink)