This is a very good question. I think the difference is between a comma-separated list of three or more subjects, and a simple ("and"-separated) two-part compound subject.
Let's take the slightly longer list: "When I opened the fridge there was only a bottle of milk, some eggs, a loaf of bread, and butter." In such a list, the items form clauses in the sentence; each one could be the singular subject or object of the exact same sentence: "... there was only a bottle of milk", "...there was only a loaf of bread", etc. You could also remove either of the two inner items with no change to the rest of the sentence (the fact that I was able to add the extra item to your quoted sentence without changing anything else in that sentence demonstrates this). In this case, the first rule you stated is correct; you pick the verb conjugation that works for the first item in the list, as if it were the only one.
However, when you get down to two items, now there are no more commas. The two items, say "a bottle of milk and butter", now form a compound subject; the two items are being referred to as one entity, which is always plural. In this case, the plural verb should be used.
EDIT: Good points. Some of those sentences sound better than others:
There is further rain and strong winds forecast for the next three days. - Not bad. I think "are" works here too.
There was a loud bang and some flashes of light before flames started pouring from the windows of the house. - This does indeed work better with "was" than "were", no question
There is a bank and cash machines in the city centre. - this one grates my ears; "there are a bank and cash machines" sounds much better for some reason.
There was no water or animals anywhere in the desert. - OK, but I think "were" sounds better here as well.
The biggest question, given the above, is why there is a difference between the second and third sentences; their structure is practically identical, the only difference I can detect is tense.
In most languages indefinite articles stem from that language's word for one. For instance in French un, or in German ein, In Italian and Spanish uno or in Portuguese um.
English is no exception: an was derived from one. Note that an was the original indefinite article; the shorter a came later when the final "n" was dropped before consonants.
In some of the languages I mentioned above, the plural form of the indefinite articles is simply formed by applying the noun plural inflection: unos/unas or uns/umas.
In others, such as German and Italian, there is no plural form to the indefinite article. Italian use the partitive article degli/delle as a substitute and this is probably also the origin of the French plural form des.
For some reason English did not go through this last step either. To understand why we need to go back to the way Old English solved the problem.
In Old English adjectives have a different declension depending on whether the noun they qualify is determined or not.
"The glad man" reads
se glæd guma
whereas, "a happy man" is:
glæda guma
As one can see, only the adjective changes.
For one given adjective, you could therefore have different inflections depending on:
- the noun gender (masculine, feminine, neuter)
- the noun being singular or plural
- the four cases (nominative, accusative, genitive, dative)
- whether the reference is definite or indefinite.
So that the same adjective would have to follow either the "definite" declension or one of three "indefinite" declensions.
þa glædan guman
Edit
<conjecture>
The theory I'm trying to check (community please feel free to edit) is that in various languages (Icelandic for a language very close to Old English or Romanian) the article is added as a suffix to the noun. Then it often "detaches" and passes in front of the noun. Icelandic is half way through for the definite article in that matter.
As for the Old English indefinite article, my conjecture is that the process never went through for a number of possible reasons:
- The "loss of inflection" of early Middle English won the race
- The plural of "an" was not easy to evolve at that time (the Romance "-s" plural had not imposed itself yet).
</conjecture>
But the need is still there, just as in any other language where a specific word emerged for the plural indefinite article. This gap is filled by placeholders such as some or a number of.
Most linguist agree that Proto Indo European did not use articles.
Latin does not have any kind of article, and Ancient Greek arguably had no indefinite article either - it was using something very much like present-day English some (τις - "a certain"). And I believe that Old German did not have any article either.
It is a very remarkable fact that articles appeared in many modern Indo European languages in a largely mutually independent yet very similar manner. My feeling is that their emergence compensates for the gradual loss of inflection in these languages. But then present-day German is a powerful counterexample...
Best Answer
A few old grammar rules
Since Few words on Many Subjects was published in 1831, English has seen quite a few changes. I don't know whether this rule was enforced at schools, but I did find another example lambasting the use of the indefinite article before many in front of a plural noun.
a/an + adjective + number + plural noun
The fact that native speakers were using the article "a" in front of adjectives and plural nouns in the 19th century, proves there is nothing new under the sun. Today the following sentences are perfectly grammatical.
In sentence 3, “a great number of” could substitute “a good many”.
(a) He weighed twenty-five stones.
(b) We spent three weeks in Greece.
(c) He had collected many books.
(d) She waited three minutes before speaking
(e) Alice had experienced two exciting years.
(f) Ted had two exhausting days.
(g) It (only) costs twenty dollars
The sentences are only grammatical without the "a" and its "adjective"; take away only one of the two components, and the sentences become ungrammatical. The indefinite article modifies the adjective with the number. There has to be a number attached to the adjective in order for the sentence to be grammatical.
The noun phrases take a plural noun and a plural verb after the singular a great /good many; or a/an + adjective + number; e.g.
In a great (or) good many people, “great” and “good” act like the adverb very, or really, they intensify the adjective many; i.e. “very many people” and “really a lot of people”.
The determiner many and a good many are listed in all the dictionaries I checked, but they offer no insights as to why this construction is acceptable.
Rogermue in the comments, suggests that a great many is derived from the German noun Menge a word meaning "multitude".
An article in Language Log has this to say on this particular construction
The article ends with an update and suggests reading two studies. The paper, A SINGULAR PLURAL, by Tania Ionin & Ora Matushansky, I believe would interest the OP a great deal.