Violet sits at the short-wave end of the visible spectrum and was for most of history the hardest region of it to reproduce with any permanence.
The distinction from purple matters here and is routinely lost. Violet is a spectral hue, a single band of wavelengths that the eye can be shown directly. Purple is not: it is a mixture, produced by stimulating the red and blue ends together, and it corresponds to no single wavelength. Newton's spectrum ends in violet and has no purple in it at all.
The pigments were poor. Mixtures of red and blue lakes shifted and faded. A stable inorganic manganese violet did not arrive until the nineteenth century, and cobalt violet was expensive. Painters before that period who wanted the colour generally glazed a red over a blue and accepted that it would not last, which is one reason surviving old paintings under-represent violet relative to what was originally there.
The decisive change came from dye chemistry rather than pigment chemistry. In 1856 William Henry Perkin, then eighteen and attempting to synthesise quinine, obtained instead a purple aniline dye he called mauveine. It was the first commercially successful synthetic organic dye, it made him wealthy, and it began the coal-tar dye industry that within decades replaced most of the natural dyestuffs described elsewhere in this atlas.
A colour that had been the mark of rank because it could not be had cheaply became, within a few years, the colour of a fashion.