A still uses about 2600MJ to process 1 bucket of biomass. It does so at a maximum of 6MJ/t, so processing a bucket of biomass into .3 buckets of biofuel takes ~433 ticks.
The fermenter takes about 1400MJ per 'operation' with a maximum input of 17MJ/t. If you fuel the fermenter with saplings, you get 0.8 buckets per operation, so it takes 1750MJ per bucket. At 17MJ/t that would take ~103 ticks to complete. The other common option is compressed plant balls, which produce 0.5 buckets per operation. This would end up taking 2800MJ/bucket, and one bucket every ~164 ticks.
The math seems to say that you need at least 4 stills to be able to keep up with a single fermenter running on saplings, and 3 to keep up to one running on compressed plant balls. This would take 17MJ/t + (4x 6MJ/t) = 41MJ/t for saplings or 17MJ/t + (3x 6MJ/t) = 35MJ/t for compressed plantballs, discounting any inefficiency losses in cabling / piping.
After some more research of my own
I decided to do some more research about energy usage for various materials, I ended up with these numbers for the fermenter, with ticks/bucket calculated:
Item MJ/item MJ/bucket of biomass Ticks/bucket
Plantballs 725 1620 95.3
Compressed plants 806 1600 94.1
Saplings 1295 1541 90.6
The still used 25,400MJ for 10 buckets of biomass, producing 3 buckets of biofuel. Calculating the speed, this would have taken 423 ticks per bucket.
I'm assuming the max MJ/t usage I used is correct, because these numbers come from the official Forestry wiki. This gives us the following info:
Item Ticks/bucket fermenter Needed stills per fermenter
Plantballs 95.3 4.44
Compressed plants 94.1 4.49
Saplings 90.9 4.65
Oddly enough, while I was testing, I noticed that the various items you can feed fermenters create slightly more than their listed buckets per item:
Item Listed buckets / item Actual buckets / item
Plantballs 0.4 0.448
Compressed plants 0.5 0.504
Saplings 0.8 0.840
I've taken the difference in output into account by measuring the output using liquiducts feeding into railcraft multiblock tanks that are accurate to the millibucket.
Another consideration is that there is some contention over the efficiency of redstone energy cells, which I used to test the energy draw of the fermenters and stills. Some claim a 5% loss for every junction (which would be 1 in each of these test scenarios, the cells are placed directly adjacent to the energy consumer). I've assumed 0 loss, partially to simplify the calculations, but also to present a common real world scenario. Most energy setups will use redstone energy cells as a buffer, so the loss (if it exists) is calculated in here. Note also that the loss would apply to all the machines, so ratios shouldn't be different.
Oddly enough, in contrast to what my calculations would seem to suggest and what Unionhawk's testing seems to suggest, 5 stills weren't able to keep up with a single fermenter fed with fertilizer and saplings, powered by a redstone energy cell. I've included a screenshot. As you can see, the indicators on the side of the fermenters show them almost half full, the buffer tank for biomass on the left is empty, biomass is flowing from the tank / fermenter to the stills, and biofuel is flowing to the storage tank to the right (img album).
I have no idea why these five stills can't keep up, contrary to Unionhawk's testing and my own calculations. It's a close matter, it took probably upwards of 10 stacks of saplings and more than 600,000MJ on the fermenter to get the stills halfway to full. Perhaps this is some kind of liquid duplication bug.
Best Answer
This depends entirely on the random seed used to generate the world.
Update: This blog post by Notch on Terrain generation in Minecraft may be of interest to you even though it doesn't give you the figures you are looking for. Basically: The terrain is generated from Perlin noise with some manipulations to ensure that there is enough land and water. This means the ratio will fall within some range (but I can't say what it is, sorry).