In the 1800s, scholars who knew a lot of languages (including, often, several modern and ancient European languages, and increasingly also Sanskrit) started to notice some amazing patterns about similarities in words among all of these languages, including languages from very far apart and that had no visible connection at all from the point of view of their present-day speakers. As these patterns were explored, they…
wódr̥ (PIE) → udōr (Proto-Italic) → unda (Latin) → undulātus (Late Latin) → undulate (English)
wódr̥ (PIE) → watōr (Proto-Germanic) → *watar (Proto-West Germanic) → wæter (Old English) → water (Middle English) → water (English)
In fact, I'm now noticing that the t in the Proto-Germanic line corresponds to the d in the Proto-Italic line, just as you said.