I almost envy you, having the wonderful journey of discovery still
almost entirely ahead of you. (That's not to say I've stopped learning
about LINQ - far from it.)

I'm assuming we have a situation like this:

public class Order
{
   public IEnumerable<Line> Lines { get; set; }
}

public class Line
{
   public int ItemNumber;
   public int Quantity { get; set; }
}

and you have an IEnumerable<Order> to examine.

You want to end up with an IEnumerable<int> which consists of all the
ItemNumbers of all the orders, but distinct. In other words, "all the
items we've sold *any* of". Is that about right?

The simplest way to do this is to use SelectMany(), as you've
suggested. That's usually more easily expressed in a query expression.
So, we'd have:

var allItems = from order in orders
              from line in order.Lines
              select line.ItemNumber;

var distinctItems = allItems.Distinct();

So, allItems effectively flattens the sequence of item numbers, and
then the second statement finds the distinct items. The query
expression boils down to this:

var allItems = orders.SelectMany (order => order.Lines,
                                 (order, line) => line.ItemNumber);

Does that help? Let me know if I've missed the boat, and I'll have
another go :)