All knowledge in the physical sciences is extracted from data generated by approved instruments and can be validated/reproduced by experts in the relevant branches of science. The proof of that knowledge, i.e. when we say 'we know' and we really do know, lies in the ability to apply that new knowledge to solve the problem that was the main reason of generating the data in the first place. To avoid any ambiguity in interpretation of data, each datapoint must have an unbroken link to the instrument used to generate it, while each instrument will have a unique symbol assigned to it. The boundaries of each dataset define the knowledge domain that can be extracted from that dataset and it follows that the history of the instrument used to measure the data, defines the knowledge about that data. For example, all our 'knowledge' about the air temperatures of our planet can be extracted from the data that are generated by calibrated thermometers and, since the first calibrated thermometer was invented and manufactured in 1714, it follows that we do NOT know anything about temperatures before 1714.