I gave it a little thought this evening, Don. The approach I've used is in the picture progression. First, I've drawn a single spline that represents the outline at the junction of the sole and body. I've used linear copies to make a network that I've lofted to describe a skin. I haven't tried to contain the skin within the boundaries of the iron's body (mimic the profile of it), because I'm using associative properties to let me edit the shape of it even when it's complete, without having to rebuild the entire thing.
Next, I've made a simple sweep of the profile of the body. It's only one millimetre deep, so I can use extrude to face from the sweep to the loft. TC has given me a counterintuitive result here; the picture shows the extrusion from the reverse side of the loft. No matter; I've made another sweep that meets it, and subtracted the earlier one.
The advantage of the approach I've used is that editing the copies of the first spline I drew edits the shape of the iron body, without having to be directly associated with it. I've used subtraction to take out the hand-grip area, and then applied some blends to the vaguely-iron-shaped solid. After each of those additional actions, editing the splines changes the shape while the blends are retained. If I propose a radical edit, after whick TC's not capable of reconstructing the parametric shape, depending on what I was editing, I might lose the blends, or my change might make the original surface unachievable, which has more radical implications. Undoing the last action takes me back to a safe version, so nothing is lost, and my next attempt will be more cautious!