A virus was filed as a bacterium because it was big enough to see

Mimivirus is about 0.4 micrometres across, visible in a light microscope, with a genome larger than some bacteria. It was stained, named as a microbe, and only later recognised as a virus that infects amoebae.

In 1992 a pneumonia outbreak in Bradford turned up what looked like a Gram-positive bacterium inside an amoeba. It was christened Bradfordcoccus and left in a freezer. A decade later Bernard La Scola and colleagues looked again and found no ribosomal genes, no cellular life, and a particle whose coat and cycle were viral. The particle is larger than many bacteria. Its genome, at the time, was the largest viral genome known, and it encodes pieces of the protein-making machinery, which viruses were not supposed to bother carrying. The amoeba is the host. The virus builds a factory inside it.

Most viruses are still a tenth of this size or smaller, and most still do nothing you could mistake for a cell. Mimivirus did not repeal the category. It showed the category had been drawn around the viruses we knew how to find, which were the small ones. Once people looked for large particles inside amoebae, they found a zoo of them. The definition bent. A virus can be bigger than a bacterium, and a bacterium can be smaller than a virus, and the microscope alone will not tell you which is which.