The American Naturalist 49: 5-21 (January 1915)
Some Fundamental Morphological Objections to the Mutation Theory of De Vries

Seeds on horseradish and Lilium candidum
Prof. Edward C. Jeffrey

Physiological sterility is frequently due to entirely different causes than genetical lack of harmony, as for example in the horseradish or the potato (Solanum). In the former it has been found possible to bring about the formation of fertile seed by simply girdling the top of the subterranean storage region of the plant, so as to prevent the undue descent of assimilates. The common white lily, Lilium candidum, presents a similar condition, for here the setting of seed takes place only when the leafy flowering axis is severed from its bulb and kept in water. So far as I am aware, there have been no experiments as to the result of severing the continuity of the phloem (girdling), in relation to the restoration of seed production in the potato. The common yellow day lily (Hemerocallis) possibly presents a case similar to that of Lilium candidum, for it does not ordinarily set seed, although in all the examples I have examined the pollen was morphologically perfect.

CybeRose note: Donald Beaton (1861) offered another method for lilies:

Mr. Knight made an experiment for getting early Potatoes to seed by planting them on a ridge, and when the plants were ready to bloom he washed away the soil of the ridge to prevent them making young tubers, and so force the whole strength of the plants or roots into the stems and foliage to see if that would force them to seed. Another form of that experiment is applicable to all bulbs and tubers which form roots on the flowering-stems, as the Japan Lilies and others do. Pot such bulbs or tubers with the neck of the bulbs just at the surface, and when the stem is an inch or two put an empty pot over it, introducing the stem through the hole at the bottom of the pot, then earth up the stem, and when it roots and fills the upper pot separate from the bulbs, then cross it.