The question might be re-phrased:
Given this short piece of canon, can we say anything about related
The basic answer to that is "yes". We can say that taking all tieflings as a group, the majority of them are at a far-removed generation from where the Fiendish taint was introduced to the bloodline. This rules out fiendish inheritence schemes where the probability of manifesting the taint (i.e. being a tiefling) falls off very rapidly after the first generation.
However, that is only a small proportion of all possible schemes you could invent.
Does that mean that the probability that any given descendant will manifest the taint grows over time?
No, because generation size grows over time, a fixed probability will produce higher numbers in each generation.
However, it doesn't rule it out, either. A scheme where the probability grew on each generation (e.g. 0.1, 0.2, 0.3, 0.4...0.9, 0.91, 0.92...) would fit canon.
Does it mean that the chance of the very first generation manifesting the taint is small?
No, again because the first generation is outnumbered by members in future generations. It is very easy to construct schemes that have high probability in generation 1, much lower in generation 2 onwards, which fit canon. For example, a probability of 1.0 in first generation and 0.25 in subsequent generations would easily result in the bulk of all tieflings being in a far-removed generation.
However, it doesn't rule it out, either. A scheme with probability 0.01 in first 3 generations, and 0.25 in all future generations would fit the canon.
In short, the quoted canon says very little about probabilities, and even if you are concerned about running a game adhering strictly to this small piece of text, then you have huge leeway on how to interpret it.
For further pedantic-ness, this answer is making an assumption that the rate of ongoing tainting is small compared to the size of current tiefling population. A large amount recent Fiendish love affairs in an adventure background would complicate things - and if you wanted to have that in a world background, plus stick to canon as written, plus remain self-consistent in a purely mathematical sense, you would indeed need to have a lower initial probability.