Why thin-walled tube beats solid bar
Where stiffness comes from, and why the middle of a bar is nearly useless.
Thin-walled tube beats a solid bar of the same weight because resistance to twisting and bending comes overwhelmingly from material placed far from the centre of the section, and a tube spends every gram of its material out at that useful radius while a solid bar wastes most of its material near the middle, where it barely contributes anything at all.
Twisting a curtain rail against a solid rod
Gripping a hollow metal curtain rail at both ends and twisting it against itself makes this obvious in the hands. The rail resists twisting with real force for how little metal it contains, because that metal is spread out at the rail's full diameter, working at maximum leverage against the twist. The same weight of metal drawn into a short, solid rod of much smaller diameter twists far more easily, despite containing exactly the same amount of material, because none of that material has been given any leverage to work with. The rail's advantage comes purely from where its material sits relative to the axis being twisted, with no help from a better alloy or a different heat treatment, which is the whole argument for tube over bar in a shape almost anyone has held.
Why the middle of a bar does so little
When a length of metal is twisted or bent, it does not resist evenly across its whole cross-section. Material near the centre line barely moves relative to its neighbours and carries almost no stress, while material near the outer surface moves the most and carries the highest stress. The leverage any point contributes grows with its distance from the centre, and it grows faster than that distance does, because a point further out both strains more and acts on a longer lever. A solid bar therefore has plenty of material near its centre doing very little useful work.
A tube of the same weight can be made with a much larger outer diameter, since removing the useless core frees up material to be moved outward, and every bit of that relocated material sits where a twisting or bending load needs resistance. This is why a chassis designer chasing stiffness for a given weight almost always reaches for a larger-diameter, thinner-walled tube before a smaller, thicker one, let alone a solid section, unless some other constraint (a bolt hole needing enough wall to bite into, or a mounting point needing local reinforcement) forces a compromise.
Double the diameter, keep the weight, get four times the stiffness
For a thin-walled tube, torsional stiffness grows with the cube of the radius multiplied by the wall thickness, while the weight grows only with the radius multiplied by the wall thickness. Doubling a tube's diameter and halving its wall keeps the weight the same, and the arithmetic then gives eight times from the larger radius and half from the thinner wall, so the new tube is about four times as stiff in twist for exactly the same mass of metal. Doubling the diameter again, and halving the wall again, would quadruple it once more.
That relationship also shows where the search for stiffness runs out. Thinning the wall too far in pursuit of a larger diameter eventually makes the tube's surface vulnerable to buckling locally, crumpling like a drinks can before the material itself is ever fully stressed, which is why practical wall thicknesses settle into a fairly narrow band for a given tube diameter and loading.
Reading a tube spec sheet
Outer diameter matters far more than wall thickness for stiffness, so saving weight by simply thinning the wall of an existing tube gives back stiffness faster than it gives back weight, while the same saving achieved by stepping up to a larger diameter and a thinner wall together can hold stiffness almost constant. Chassis tube sizing is rarely a simple lookup for the same reason, since the designer is balancing the stiffness gained from a larger diameter against buckling, weld access, and how far a given tube size can be bent around a driver's cockpit without kinking, all of which push back against diameter growing without limit even though stiffness alone would reward it endlessly.
Where a thick wall earns its weight
The tube-over-bar argument weakens wherever a section has to do something other than resist a clean twist or bend along its length. A short stub of tube welded into a corner bracket, carrying a bolt that clamps through it, needs enough solid wall to resist the bolt crushing or ovalising the hole, a job a very thin wall does badly however well it would resist a pure twist further along the same tube. A mounting point that takes a sharp, concentrated load from a suspension rocker likewise often needs a local insert or a thicker-walled section grafted in, because the load there is concentrated at one spot around the circumference instead of spread evenly the way a clean torsional load is, and a thin wall can crush or dent locally long before the tube as a whole would ever twist.
A real chassis frame is rarely built from one uniform tube size throughout for this reason. It uses thin-walled tube for the long runs where pure stiffness is the goal, and locally reinforced or thicker sections wherever a load has to be introduced into the structure at a single point, which is the curtain rail's lesson applied with some care about where the rail gets screwed to the wall.