Okay, Neeners, you hollered for the dog, so here ya go with the bones I've dug up....
Here are some possible whys, hows, and wherefores:
****
Some people call this "glycerin rivers", but that is a bit of a misnomer, since a true glycerine river would be soft and goopy. Some call it stearic streaking, but there's more than one fatty acid involved, not just stearic. Probably the most accurate name is just "streaking" or "mottling".
Soap is made of many different kinds of soap molecules. As the soap cools in the mold, some parts of the soap may harden before other areas do. That affects where the colorant ends up. Also some soaps are clear and others are opaque. This creates the streaks or mottles -- areas of clear soap surrounded by colored or opaque soap.
Any soap can streak, but slow cooling or gelling, titanium dioxide, and palm oil seem to increase the chance of streaking/mottling. Soap cool. Avoid gel. Put your powdered colorants in glycerin, oil, or water and let them stand for several hours before use so they get thoroughly wetted. Mix palm oil well before measuring it.
I also suspect the way the soap is put into and manipulated in the mold is also a factor, since the streaks are often aligned with the way the soap has been spooned or poured in and/or swirled in the mold. In soaps with this issue, the streaks tend to be parallel to the path the soap took when it was poured into the mold when it was poured or along the path created when it was swirled. Not sure how to avoid this technique-related streaking. My guess is that batter that is especially thin when it's poured and swirled is more prone to streaking than batter that is poured and swirled when at thicker consistency.
****
The old soap makers (1800s to early 1900s) made mottled soaps that were popular with the customers of the day. They would add a coloring agent to a finished soap, pour the soap into "frames" (large molds), and carefully control the rate of cooling.
The stearic soaps would solidify into pale colored clumps first, essentially concentrating the color into the remaining fluid oleic soap. When the oleic soaps solidified, the color would be trapped within these areas and make rivers or veins of darker color around the stearic clumps. The size and appearance of the mottles were controlled by the oils in the recipe, the way the finished soap was handled, and the rate of cooling in the frames.
****
I gleaned these thoughts from other sources. They might give you some useful tips:
using palm oil that is not homogenized (i.e.: some batches would get an unusually high stearic fatty acid content and be more prone to mottling)
using a higher water content in the soap batter than is necessary
the recipe has an unusually high amount of palm oil - a recipe with palm at 70% didn't work, but worked fine at 40%
Source:
http://health.groups.yahoo.com/group/Bath_and_Body/message/17199
soap gets warm and fluid enough during gel so it can move around and form veins or mottles
soap was taken to a full, thick trace and the "stir lines" created in the soap from your spoon/spatula can create areas of clear soap
soap was taken to a full, thick trace, and then colorant or FO was added, but the colorant or FO did not fully mix into the thick soap batter, even with vigorous mixing
Source:
http://www.millersoap.com/botched2.html Search for "mottled soap"
"...When [soap] is permitted to cool rapidly the colouring matter remains uniformly disseminated throughout the mass; but when means are taken to cause the soap to cool and solidify slowly a segregation takes place: the stearate and palmitate form a semicrystalline solid, while the oleate, solidifying more slowly, comes by itself into translucent veins, in which the greater part of the coloured matter is drawn. In this way curd, mottled or marbled soap is formed..."
Source:
http://www.1911encyclopedia.org/Soap
stearic soap streaks from nonhomogenized palm
too much titanium dioxide
soaping under 100 deg F
using a recipe that has a high stearic content
soap at 110 - 115 F when using palm
Source:
http://www.soapmakingforum.com/f22/cracklemottledcoldprocesssoap27681/