Numerals versus words, Part IV: When to make exceptions

It's time to wrap up this series with a quick look at why the rules alone are sometimes a bit insufficient. At these times, a less strict judgment is in order.

Now don't be put off by this idea. After spending three previous posts (The Basics, Context, and Names) learning the rules, we're not throwing them out the window. Those rules are enough to answer the vast majority of your questions about whether to use a word or an Arabic numeral. Just be prepared for those odd times that the nature of the content demands a little different approach.

This situation often happens when more than one figure of the same type is quoted, but some of them fall on a different side of the rule from the others. Let's use a hypothetical medical study to demonstrate:

In the first round of clinical testing, only 8 of the 200 participants responded positively to the drug. But in the second round, which differed only by a modest increase in dosage, 151 participants were completely cured of all symptoms while the remaining 49 showed markedly decreased symptoms.

In the case above, a rule is bent in order to achieve consistency, appropriate emphasis, and easy readability. If a very strict interpretation of the rule were followed instead, the numbers (8, 200, 151, and 49) would be harder to find, compare, and interpret because some of them would be lost among other words:

(according to Chicago Manual of Style's rule to spell out numbers under 100)
In the first round of clinical testing, only eight of the 200 participants responded positively to the drug. But in the second round, which differed only by a modest increase in dosage, 151 participants were completely cured of all symptoms while the remaining forty-nine showed markedly decreased symptoms.

(according to AP Stylebook's rule to spell out numbers under 10)
In the first round of clinical testing, only eight of the 200 participants responded positively to the drug. But in the second round, which differed only by a modest increase in dosage, 151 participants were completely cured of all symptoms while the remaining 49 showed markedly decreased symptoms. 

Again, these kinds of judgment calls will be rare for many writers. Unless you deal with lots of numbers in text, you may not have to go beyond the rules at all. But if you find it necessary to wiggle the rules in order to create a clearer representation of your ideas, don't feel bad about it. All major style guides acknowledge the necessity for exceptions at those times.

Readers, chime in. Where have you seen exceptions to numbers rules exercised? Also, as this is the last planned installment of the numbers series, please speak up if there's a finer number question you'd like to see explained in full instead of glossed over. I will certainly revisit the topic at your request.