Estimating Cell Count

I'm pondering washing the yeast from a batch that I'm bottling tonight. I intend to make a starter with it for my next batch. But my question is: with yeast harvested from a given fermentation, and then run through a starter, how do you know with any kind of accuracy how many cells you are working with, and therefore if you're overpitching/underpitching to the next batch ? Or is it more of a try it, and if it works, repeat... type of process ?

to really know you need some specialized equipment.
A microscope, and there is a special plate to put the yeast sample on, which puts a grid in the view for you, then you put on a measured amount of yeast slury - something like a drop, but it is very specific in size, like a micro liter (.001 ml) - ok, I probably have that size wrong. with it on the grid, you can count the cells. After geting a count (1500) in .001 ml, then you know you have 1500,000 per ml etc etc.

After that I've got now idea how to figure it from a batch of starter by eyeballing it. A brewer at work makes a judgement by the kerousan ... if he gets a high one for that style yeast, he repitches, if it is low, he buys new.

Well the answer is...you don't know with any decent accuracy, really. But then again, the amount of yeast needed for a batch is also known only to a broad estimation.

The best approach is probably to plug the amount of slurry you have into the MrMalty yeast calculator for slurry, making your best guess as to the parameters like yeast% and such. Then use that number in your starter calc either on MrMalty or YeastCalc.com. And then maybe adjust your starter amounts upward a bit, since it's generally accepted that it's better to overpitch a bit than to underpitch.

But really, without instruments it's a seat-of-the-pants type situation.

I think I read recently that one of the labs says a standard slurry of 40-60% yeast will have 1.2 billion per ml. Then you can calculate a loss in viability as it gets older. You should check out woodland brewing blog; he has viability numbers. Generally storage is better than the yeast calcs will indicate.