Sir

Your News Q&A article 'Lab disinfectant harms mouse fertility' (Nature 453, 964; 2008)

must have set tongues wagging in coffee rooms throughout academia. We experienced a similar catastrophe, which took two years and exhaustive detective work to resolve, simply because we moved laboratories.

Our research relies on a widely used model of allergic lung inflammation, in which mice are exposed to a model allergen. While we were working at the University of Cambridge, the model was always reproducible. But when we returned to Ireland to continue this research, we found the lung physiology of the control mice was inexplicably abnormal for the first year.

After months of revising protocols, testing reagents and pathogen screening, we noticed that all the mice developed spontaneous, non-specific lung inflammation within four weeks of arrival, indicating that an environmental factor was probably to blame.

To cut a long story short, it turned out that mouse chow sterilized by steam autoclaving caused the release of fine particulates, and these were inhaled by mice in their individually ventilated cages. Mice fed instead with irradiated chow had normal lungs.

Unless peer-reviewed and published, such discoveries become anecdotal. So why did we not publish these findings? After a two-year hiatus doing experiments outside the area of our core research, the pressure was on to focus once more on our research aims. Additionally, the growing emphasis on commercially exploitable research, rather than on basic science, means that funding bodies are not impressed by publications on empirical investigations.

This raises a broader, unspoken question in the field of mouse immunology. Why are there differences between data generated by different groups working on apparently the same mouse model or strain?