I observed one night a candle burning brightly, then dimming, then building again and burning brightly, with a period of about 30 seconds. I observed the oscillations for a few minutes and thought to myself: Delayed Feedback.
Not all candles do that, so why did this one? The wick had folded over on itself such that the volume of the wick compared to its evaporative surface was high. Additionally, the cross sectional area of the wick at the level of the wax puddle was still rather low.
My hypothesis is that the small cross sectional area of the wick meant that it could only soak up so much wax at a time, and so constituted a delay. The folded top of the wick was a reservoir that was involved in a positive feedback - the hotter the flame, the more wax evaporated from it, and the hotter the flame burned. When the reservoir in the wick was depleted, it needed to wait until the wicking action recharged it.