A solarization occurs when the film and/or paper is exposed to light during development giving both negative and positive qualities in the same image. (For instructions on how to simulate a solarization in Photoshop, click here.)
Because it is very difficult to solarize film (and because I only know how to do it with a discontinued film [Kodak Panatomic-X]) I will restrict this discussion to solarizing paper.
The trick is NOT to expose during development but to re-expose and re-develop. In order to do this, one must suspend development rather than stopping it. What is needed is an intermediate solution of IODIZED salt and water - which is used like stop bath.
The difference is that stop bath uses acid which makes the paper no longer responsive to developer. The IODIZED salt stops development without rendering the paper unresponsive to development (It requires a large amount of the IODIZED salt: 1-3 pounds for a 16x20 tray)
Contrast: the re-exposure causes a flattening effect (this is worse when one just flashes the paper with light while it is in the developer) so you must compensate. Start with a high contrast negative. Use high contrast paper (I used Agfa #5 and Oriental #4) Use high contrast developer (I used undiluted Kodak Dektol with a pinch of hydroquinone.)
Expose the paper as if printing only the shadow areas of the negative. Let the "blocked up" highlights wash out. Develop fully. Submerse the print in the in solution of IODIZED salt and water and agitate vigorously for 30 seconds.
Place the paper on a tray and return to the enlarger (it is best if you have more than one enlarger) and re-expose the paper to pure white light (no negative). Go back to the developer and redevelop fully. Then stop, fix, clear and wash normally. The exposure times will require experimentation.
You might try the Sabatier method of making a test print, but it never worked well for me.
Solarizations are frustrating exercises in trial and error. Expect it.
Feel free to contact me with questions/problems. I will answer if I am able. There is also a good article here.