Pulling a bar until it breaks
In guide 1 you learned the language: stress is force spread over area, strain is the fractional stretch it causes. The tensile test is the machine that speaks that language out loud. You clamp a standardized 'dogbone' specimen — fat gripping ends, a slim uniform middle — into two jaws, and a screw or hydraulic ram pulls them apart at a slow, steady rate. A load cell reads the force at every instant, and a little clip-on gauge called an extensometer reads exactly how much the slim middle (the 'gauge length') is stretching. It is the single most common test in all of materials engineering, and almost every property number you will ever quote for a metal starts here.
The raw numbers are force and elongation, but those depend on how thick and how long your particular bar happened to be. So, exactly as in guide 1, we normalize. Divide force by the original cross-section area to get engineering stress, and divide the stretch by the original gauge length to get engineering strain. Concretely: a round bar 10 mm across has area A0 = (pi/4) times (10 mm)^2 = 78.5 mm^2. Pull it with 15.7 kN and the stress is 15700 N / 78.5 mm^2 = 200 MPa (remember 1 MPa is just 1 N/mm^2). If its 50 mm gauge length stretches by 0.05 mm, the strain is 0.05 / 50 = 0.001. Now the numbers describe the steel, not the bar.
The whole curve at a glance
ENGINEERING STRESS-STRAIN CURVE (a ductile metal) stress | UTS | __.__ | /' '~-._ <- necking | /' 'x <- fracture | / plastic flow | yield o | /: (0.2% offset yield) | / : | / : <- elastic: straight line, slope = E (Hooke) | / : +---+----+---------------------------> strain 0 0.002
Before we zoom in, appreciate how much this one picture carries. The slope of the straight part gives the stiffness. The knee where it bends over gives the strength you can use without permanent damage. The peak gives the maximum load it will ever carry. The strain where the line finally stops gives how much it stretched before dying. And the whole shaded area underneath gives the energy it soaked up. One test, five very different design questions answered — and, crucially, they do not move together, which is where most beginner intuition goes wrong.
The elastic part: springy, reversible, and set by bonding
The first stretch of the curve is a straight line, and it is the friendly, forgiving part. This is elastic deformation: pull within it and the bar springs perfectly back to its original length the instant you let go, like a stiff spring. The straightness is Hooke's law — stress is simply proportional to strain, stress = E times strain — and the slope of that line, E, is the Young's modulus, the material's stiffness. It measures how hard you must pull to get a given stretch, and physically it is just the atomic bonds being stretched a hair and pulling straight back, so E is set by the bonding-energy curve you met in the bonding rung.
Numbers make this vivid. Steel's Young's modulus is about 200 GPa; aluminum's is about 70 GPa. So at the same 200 MPa stress, steel strains only 200/200000 = 0.001, while aluminum strains 200/70000 = 0.0029 — nearly three times as much give for the same push. Here is the honesty that trips people up: Young's modulus is almost immovable. Anneal your steel, quench it, alloy it, cold-work it — its yield strength can swing from 250 MPa to over 1500 MPa, but its stiffness stays glued near 200 GPa, because you have not changed the bonds, only the defects that block sliding. Stiffness and strength live in separate boxes. Guide 3 unpacks the elastic region in full.
Two companions ride along in this elastic zone. Pull a bar longer and it also gets thinner sideways; the ratio of that sideways shrink to the lengthwise stretch is Poisson's ratio, about 0.3 for most metals — so our 0.001 axial strain comes with roughly 0.0003 of lateral contraction. If instead you twist or shear the material, its resistance is the shear modulus G (about 77 GPa for steel), tied to E and Poisson's ratio by G = E / (2 times (1 + v)). And the triangular area under just the elastic line is the resilience — the elastic energy the material can store and give back like a spring, which is why a good spring steel has a high yield strength and a modest modulus.
The plastic part: permanent, one-way, and the peak
Pull harder and the line bends over. You have passed yield: dislocations start gliding (the slip of guide 1), and now some of the stretch is plastic deformation — permanent, one-way, still there after you let go. Where exactly does elastic end and plastic begin? Real metals bend over so gradually that the point is genuinely fuzzy, so engineers use a rugged, reproducible convention for the yield strength: the 0.2 percent offset.
- Find the straight elastic line and note its slope (that is E).
- Go along the strain axis to 0.002 — that is 0.2 percent permanent stretch, the small amount we agree to tolerate.
- From that point draw a line parallel to the elastic slope.
- Where that offset line crosses the curve, read off the stress — that is the 0.2 percent offset yield strength.
Past yield the curve keeps rising, because plastic flow tangles dislocations and makes the metal harder to push (work hardening). It climbs to a single high point, the ultimate tensile strength (UTS) — for a mild steel, maybe 400 MPa above a 250 MPa yield. Here is a stubborn misconception to kill now: the UTS is not the stress at which the bar breaks. It is only the highest point the engineering curve reaches. A ductile metal sails past its UTS and breaks later, at a lower engineering stress, after it starts to neck.
At the peak the stretch stops spreading evenly and localizes: one spot thins into a neck, all further deformation crowds into it, and the bar quickly tears there. How much did it stretch before dying is its ductility, measured two ways — percent elongation (how much longer the gauge length got, roughly 25 percent for mild steel, under 1 percent for grey cast iron) and reduction of area (how much the neck shrank). And the total energy soaked up all the way to fracture — the entire area under the curve — is the toughness. Guide 4 gives yield, strength, and ductility their own full treatment.
Engineering vs true stress-strain: the honest correction
Now for a puzzle the curve hands you. After the UTS the engineering stress falls — the line droops down to fracture. Does that mean the metal is getting weaker as it necks? No, and believing so is a classic trap. Remember the note from the start: engineering stress keeps dividing force by the original area A0, but the neck's real cross-section is shrinking fast. Divide a falling-but-still-large force by a stubbornly-fixed A0 and the number drops — even though the metal in the neck is still getting stronger.
The honest fix is true stress-strain: divide the force by the instantaneous area (true stress = F / Ai) and use the natural log of the length ratio for strain (true strain = ln(Li / L0)). Plot that, and the curve does not droop — it keeps rising all the way to the break, because work hardening never actually stops; the metal really is strongest just before it fails. So the engineering droop is an artifact of our bookkeeping, not real softening. We still tabulate and design with engineering values, because in practice you rarely know the instantaneous neck area — just never read the droop as weakness.
Hardness, and designing with the scatter
A full tensile test destroys a machined specimen and ties up a machine, so for a quick check we often measure hardness instead: press a hard indenter into the surface with a known load and measure the dent. A big ball gives Brinell hardness (HB); a cone or ball measuring depth on a dial gives Rockwell hardness; a tiny diamond pyramid gives Vickers hardness, fine enough to probe a single grain. All three measure resistance to local plastic flow, so they track strength: for steels the rule of thumb is TS (in MPa) is about 3.45 times HB, so a Brinell 120 steel runs near 414 MPa. Honesty flag: hardness is not hardenability — hardenability is how deep a steel hardens on quenching, a different idea you meet in the heat-treatment rung.
Step back and hold the guide's central lesson: stiffness, strength, and toughness are three genuinely different things, and the curve shows why. A ceramic gives a tall, straight, short line — very stiff and strong, but with almost no area under it, so it is brittle: low toughness. Annealed copper gives a low but enormously long line — soft and weak, yet a huge area, so it is superbly tough. Neither is 'better'; they are different. Collapsing all three into a single word 'good' is the most common beginner error, and this curve is your cure for it.