In Nature in 1939 he predicted that the structure of insulin would be solved in this way - a task finally accomplished by Dorothy Hodgkin and her Oxford group in 1969. However, it was the determination of the structure of limonin by the Robertson group in 1961, which first proved the feasibility of the method.
The use of X-rays to determine molecular structures had previously been limited to relatively simple structures due to the inability of scientists to measure the phases of X-ray waves. Robertson solved the "phase problem" by realizing that if he could replace one (or more) of the atoms in a molecule with heavier alternatives, without changing the molecular shape (isomorphous replacement), then the enhanced X-ray scattering from the greater electron density in the heavy atoms would allow him to measure the phases directly.
This breakthrough signalled the start of a revolution in chemical crystallography and structural biology, leading to the many thousands of structures of natural products, synthetic compounds, proteins and other biological macromolecules that we know today.
The 1961 paper is also remarkable as an example of the early use of the newly-developing programmable electronic computers to perform some of the tedious mathematical manipulations required to calculate structures from X-ray diffraction data.