I was wondering why there is a "th" in mother, father, and brother, but not in sister? Is the etymology of the word different?
The short answer is yes; sister is formed a bit differently from the other words you list for as far back as we can trace. However, the evolution of these words over time has been a little complicated. Since the word "daughter" is also related, I'll discuss it as well.
As John Lawler says in his answer, all of these words are inherited from Proto-Indo-European (and that's the furthest back that we can currently trace them with any certainty). The Proto-Indo-European (PIE) roots are reconstructed as follows, more or less (for the ones with less consensus on the details, I include several variant reconstructions separated by slashes):
You can see that at this point in time, the last root, for "sister," only has the final consonant in common with the others. On the other hand, the first four all seem to share the same ending (tēr).
However, PIE did have some kind of stress or pitch accent, and the placement of the stress (marked with an acute accent ´) was apparently different in the different words. This will be significant to later developments! I haven't marked the stress on the "mother" word because its descendants in different languages give conflicting information about the position of the stress. But in the ancestor of English, it seems the stress was on the final syllable (*mātḗr).
English is descended from PIE through Proto-Germanic (PG), which is defined as the common ancestor of all the Germanic languages. Here are the forms of the PG roots:
The relevant sound-changes that get us from the PIE consonants to the PG ones are:
- Grimm's law: causes many consonants to change between PIE and Proto-Germanic. It's why PIE *p in the "father" word became PG *f, PIE *bh in the "brother" word became PG *b, and PIE *dh in the "daughter" word became PG *d. And by Grimm's law, PIE *t became PG *th, which explains the middle sound of a single word among these, brother.
- None of the other words here have *th in Proto-Germanic because Grimm's law had several regular exceptions. One common one is that PIE *t did not turn into *th when it came after certain other consonants in PG (namely, the fricatives); this is why the "daughter" word in Proto-Germanic has *ht rather than something like *hth. PIE *st was also preserved as PG *st. However, the "sister" word didn't actually have a *t in PIE, so it's not clear if Grimm's law is relevant there. Another important regular exception is when consonants were subject to the following rule:
- Verner's law: the relevant part here is that when PIE *t came after an unstressed vowel in PIE (including a syllabic "laryngeal" like h2/ə), it ended up as *d in Proto-Germanic. This applies to the "mother" and "father" words... we'll see how they got to their present English forms later.
- Epenthetic insertion of *t to separate the PIE consonant cluster *sr (two other examples of this are PG *thimstra "darkness" from PIE *thimsra, and PG *straumaz "stream" from PIE *srow-mó-s; I took these from Piotr Gąsiorowski's paper The Germanic Reflexes of PIE *-sr in the context of Verner's Law). This process also occurred in the ancestor of the Slavic languages, Proto-Slavic. Gąsiorowski says that this sound change is "generally considered to be uncontroversial" (although he also argues that in some cases, Verner's Law regularly led to an alternative outcome of r or rr).
Note that Verner's Law means that the "brother" word actually has a different middle consonant from the "mother" and "father" words at this stage! It still does in standard High German (where we have Mutter, Vater, Bruder, Schwester) and it did in Old English (where we have mōdor, fæder, brōþor/brōðor, sweostor).
But in Middle English, there was another sound change that turned the d-sound in the "mother" and "father" words into a th-sound. (Other words affected by this change include gather, together, weather, and heather; see this question and its answers for more information: /ð/ → /d/ shift in English)
And that's how we get to Modern English
The reason there's a TH in the English words father, mother, and brother, but not in sister,
is that there was a *t in the Proto-Indo-European roots for father, mother, and brother,
but not in the PIE root for sister.
The series of consonant changes recorded in Grimm's Law made those PIE *t stops change into fricatives; these fricatives are the sounds spelled with TH in modern English. Thus,
- PIE *bhrāter- becomes ModE brother (PIE *bh becomes ModE b, also by Grimm's Law)
- PIE *pəter- becomes ModE father (PIE *p becomes ModE f, also by Grimm's Law)
- PIE *māter- becomes ModE mother
- PIE *swesor- becomes ModE sister (the additional /t/ doesn't come from PIE)
Some family trees of other English words that come in various ways from PIE roots:
I beg to differ with John Lawler’s answer. The proto-Germanic word for sister is *swestēr. So it is perfectly reasonable to ask why the /t/ in the ancestors of the words for “father”, “brother”, “mother” becomes th, while the /t/ in the Germanic ancestor of “sister” does not. The answer is that Grimm’s law affects freestanding /t/, but does not affect the /t/ in the cluster /st/.
I quote Etymonline for 'mother (n.1)':
[b]ased ultimately on the baby-talk form *mā- (2); with the kinship term >suffix *-ter-
So (and while I am no expert) it appears that originally they had 'ter' rather than 'ther.' In any case, then you'd have pater, moter, sister and broter. Then (from what everyone else is saying) Grimm's law would change the sounds of the other ones.
However, I do know (thanks to http://www.lexilogos.com/english/latin_dictionary.htm's Notre Dame link) that pater is the latin word for father in and of itself. Mater (not moter) was the latin word for mother, but the sounds are pretty similar. Sister and brother in latin weren't parallel to English, though. I can't really help there.