The theory of the constitution of atoms and molecules that Niels Bohr proposed a century ago was not motivated by specific empirical problems. Bohr was a theorist, not an experimenter. Nonetheless, he was much occupied, even obsessed with problems of an experimental nature and with confronting his theory to experimental tests. Not only did he follow relevant experiments closely, for a period he also engaged in spectroscopic and other experimental work himself. The development of the Bohr theory (or the Bohr-Sommerfeld theory), illustrates the fertility of a close connection between experiment and theory, and it also illustrates the complex role played by empirical successes and anomalies in theory testing.

As is clearly shown by the declining phase of the old quantum theory, an anomaly is not just an anomaly. Anomalies may be given different weights, and not always for good reasons. Nor is a confirmation just a confirmation, witness that some of the most impressive confirmations of the theory (such as the Stern-Gerlach effect and the fine-structure spectrum) turned out to be spurious. While some anomalies were taken very seriously, others were more or less ignored, and others again were only recognized as anomalies post factum. In my presentation I argue that (i) to appreciate the status of quantum atomic theory in early 1925, one has to take into considerations not only the problems of the theory but also its successes; (ii) one has to extend and differentiate the list of anomalies as well as confirmations; (iii) the observability criterion was of much less importance to Heisenberg’s Umdeutung than the correspondence principle; (iv) the sense of crisis in the physics community, and the revolutionary nature of the passage from the old to the new quantum theory, has often been exaggerated.