I know almost nothing about buying/selling stock and brokers. (With two exceptions [1], all my investments have been the "buy shares of a mutual fund and hold it forever except for occasional exchanges with other funds I'm going to hold forever, or when I need to make a big purchase like a new car" kind of investments). Thus, I'm quite confused here.

What is this "deposit" in this case?

Could someone walk through what happens in this case: I have $X at my cash account at a broker, and I direct that broker to buy something that will cost $Y (including any broker fees, commissions, required tax withholding, etc.), where Y <= X.

I presume that ultimately I end up wit the thing I wanted to buy, and my cash account ends up with $(X-Y)?

So what happens as a step along the road to making that happen that requires the broker to use money that is not money from my cash account?

[1] In 1998 or 1999, the newsreader I used on Red Hat started corrupting my .newsrc after an update. I tracked it down to a change in one of the ctype macros in the C standard library. The newsreader assumed that all the flags fit in a char, which had been true before the update, but now wasn't. I submitted a patch that changed one declaration in the newsreader from "char" to "unsigned short". When Red Hat was going to IPO, I was surprised to get an email telling me that since I was a contributor to Red Hat, I could buy Red Hat stock at the IPO price. I bought a few thousand dollars worth that morning at $14/share, sold them that afternoon at something like $70/share, transferred the money out, and never touched my ETrade account again.

Who says submitting patches to open source doesn't pay? :-)

The other time was when T-Mobile gave every customer one share of stock. Eventually they got tired of having to deal with having a bazillion people who owned one share each of their stock, and told us we had to transfer our shares to the broker of our choice or sell. I sold.

To me it seems strange that that isn't allowed.