Why can't you attach a marker directly to a primary antibody. Why need a secondary antibody?

Answer 1

You can attach the detectable tag to the primary antibody. But, you can buy the tag attached to an antibody to antibodies from another organism. For example, one of the cheapest antibodies you can buy is fluorescein labeled rabbit antibody against mouse antibodies. It will recognize any mouse antibody and bind this fluorescent tag to it. If you have any mouse monoclonal antibody (almost any, let's not go there), then this antibody will bind to it. Now, you only have to prepare a mouse antibody to your protein or whatever and you can detect it. No reason you couldn't bind FITC to it to make it fluorescent, but it's another lab procedure. So simplicity and price are important reasons. But also, if you use a secondary antibody where several molecules of that fluorescent antibody can bind to one molecule of your primary antibody, then you can get more signal for each antibody bound using the secondary antibody labeling.

Answer 2

Antibodies are highly variable in shape on their "business end," so you would have to construct a different marker for each one. Much simpler to have a secondary antibody which detects all antibodies.