The Save As clearly outputs single precision floats. Try putting the input values into e.g. https://www.h-schmidt.net/FloatConverter/IEEE754.html https://www.h

The Save As clearly outputs single precision floats. Try putting the input values into e.g. https://www.h-schmidt.net/FloatConverter/IEEE754.html https://www.h-schmidt.net/FloatConverter/IEEE754.html - the outputs are just the closest representable float numbers (converted back to a reasonable base-10 representation).