Two triangles are similar when every pair of corresponding angles is equal and every pair of corresponding sides is in the same ratio. You only need to verify half of that: either two angles, or two sides and the included angle, or all three sides in proportion. The three tests come from Euclid's work on geometry and haven't changed since β€” which is why this method works identically in a ninth-grade classroom, a drafting office, and a computer-graphics pipeline.

Understanding why the tests are shorter than the definition matters more than memorizing them. Once you see that the third angle is forced and that an angle between two proportional sides locks the shape, the tests stop looking like arbitrary rules and start looking like the minimum evidence a judge needs.

What Similarity Actually Claims About Two Triangles

Similarity is a shape statement, not a size statement. Two similar triangles have identical angle sets, and corresponding sides differ by one multiplier β€” the scale factor. If triangle ABC is similar to triangle DEF with a scale factor of 3, then AB/DE = BC/EF = CA/FD = 3, and every angle in the first triangle matches the corresponding angle in the second.

That multiplier applies to more than side lengths. Perimeters scale by the same factor. Areas scale by the square of it. A triangle with a scale factor of 3 next to its original has a perimeter three times longer and an area nine times larger. This is why maps, scale models, and screen resolution conversions all lean on similarity β€” the shape survives the stretch, and the measurements convert cleanly.

Do not confuse similar with congruent. Congruent triangles are similar with a scale factor of exactly 1, so congruence is a narrower case. Two triangles can share all three angles and differ wildly in size; that's similarity without congruence. The distinction matters when a problem asks you to prove one thing and you deliver proof of the other.

Notation matters here, too. The order of letters in a similarity statement encodes the correspondence: writing Ξ”ABC ~ Ξ”DEF states that angle A equals angle D, angle B equals angle E, and angle C equals angle F. Get the order wrong and you've asserted a false claim, even if the triangles really are similar. Reinforced concrete beams, surveyor triangulation, and the triangle inequality theorem in vector math make their claims in exactly this ordered form.

The Angle-Angle Shortcut: Why Two Angles Beat Three

The AA (Angle-Angle) test says that if two angles of one triangle equal two angles of another, the triangles are similar. You do not need to check the third. The interior angles of any planar triangle sum to 180 degrees, so once two angles are fixed, the third is arithmetic β€” subtract the two known measures from 180 and the remaining measure is forced in both triangles.

This is why AA is the test most people actually use. In textbook problems, geometry proofs, and practical field work, angles are cheaper to obtain than side lengths: a protractor reads them directly, a theodolite measures them at range, and parallel lines drawn by a straightedge create them. Once two angles match, the correspondence is automatic and the similarity statement writes itself.

Work the deduction explicitly so it becomes automatic. If angle A = 40 degrees and angle B = 60 degrees, then angle C = 80 degrees. If angle D = 40 degrees and angle E = 60 degrees in another triangle, then angle F = 80 degrees. Two matched angles produced three matched angles without any additional measurement. Write the third angle into your proof, even though AA allows you to skip it β€” teachers award partial credit for showing the arithmetic.

One boundary: AA assumes planar Euclidean geometry. On a sphere, the angles of a triangle sum to more than 180 degrees β€” a textbook triangle drawn on a globe with two right angles still leaves room for a third angle, which is why the third-angle shortcut collapses on curved surfaces. In geodesy and spherical navigation, you must check all three angles. The straight-line classroom geometry most readers meet operates on the plane, so AA holds.

The SAS Test: Two Sides and the Angle Between Them

The SAS (Side-Angle-Side) test for similarity says that if two sides of one triangle are proportional to two sides of another, and the included angle β€” the angle between those two sides β€” is equal, the triangles are similar. Proportional sides without the included angle equal is not enough. The angle must sit between the two sides you measured, and it must match in the corresponding position.

Consider triangle ABC with AB = 6, BC = 9, and angle B = 50 degrees. Consider triangle DEF with DE = 4, EF = 6, and angle E = 50 degrees. Then AB/DE = 6/4 = 1.5, and BC/EF = 9/6 = 1.5. Both ratios equal 1.5, and the included angles match at 50 degrees. By SAS similarity, Ξ”ABC ~ Ξ”DEF with a scale factor of 1.5.

Now break it. Change the included angle in the second triangle to 60 degrees. The two side ratios still equal 1.5, but the triangles are not similar β€” the second triangle's shape is different, because the included angle determines whether the third side is long or short relative to the two sides you measured. This is the classic mistake: treating side ratios as sufficient when the angle is missing or wrong. SSA β€” two sides and a non-included angle β€” does not prove similarity, just as it doesn't prove congruence.

SAS is the practical test in surveying and mechanical design, where you impose a distance, a distance, and the angle between them. It is also the test computer-graphics code uses when transforming one panel of a 3D model to another. If you're working through a proof and you have two side ratios plus an angle, confirm that the angle is included before invoking the test.

The SSS Test: All Three Sides in Proportion

The SSS (Side-Side-Side) test says that if all three pairs of corresponding sides are proportional, the triangles are similar. Side lengths are enough β€” no angle measurement required β€” because the angles follow from the side lengths. For a fixed set of three side lengths, the triangle's shape is rigid: only one triangle exists up to reflection and rotation, so the angles are determined even though you never measured them.

Test the ratios carefully. Suppose triangle 1 has sides 6, 8, and 12, and triangle 2 has sides 9, 12, and 18. Then 6/9 = 2/3, and 8/12 = 2/3, and 12/18 = 2/3. All three ratios agree at 2/3, so the triangles are similar with a scale factor of 2/3 (or 3/2, depending on direction). The corresponding sides must align in the correct order β€” pair the shortest side with the shortest side, and so on β€” which is why ordering the sides from smallest to largest before comparing ratios is the reliable method.

A trap worth naming: some side triples look close but fail one ratio. Suppose triangle 1 is 3, 4, 5 and triangle 2 is 6, 8, 11. The first two ratios equal 2, but the third is 11/5 = 2.2. The triangles are not similar. An old rule from the Pythagorean world β€” that 3-4-5 and 6-8-10 are siblings β€” tempts students into assuming that roughly matching numbers create similarity. Verify every ratio before asserting the test.

SSS similarity is the practical workhorse when you can measure distances and nothing else: mirror surveying, sport field marking, or checking that a pattern piece was scaled correctly on a copier. The test gives you a yes or no without a protractor.

Side-by-Side: Which Test Fits Your Evidence

The three tests take different evidence and produce the same verdict. Use the table to pick the test that matches what you can measure.

Test What you must know What you do not need Typical setting Failure trigger
AA Two matching angles Any side length Textbook proofs, angle-only sketches, classroom exercises Assuming all three angles without checking the third in a curved-space problem
SAS Two proportional side pairs and the included angle The third side or the other two angles Surveying, mechanical assembly, graphics scaling Angle not between the measured sides, or non-included angle used
SSS All three side pairs in the same ratio Any angle measurement Distance-only measurement, copier scale checks, pattern sizing Ratios differ in one pair; ordering of sides mismatched
SAS with non-included angle Two sides and an angle not between them Nothing β€” insufficient Never valid for similarity Invoking SAS with an angle not adjacent to both sides

Read the verdicts as split rules. If you have two angles, use AA and stop. If you have two side pairs and the angle between them, use SAS. If you have three side pairs, use SSS. If you have two side pairs and a non-included angle, the evidence is insufficient β€” remeasure, or find a different pair.

The table also exposes a common decision error: spending time hunting for the third side ratio when an angle measurement you already have would close the proof immediately. Choose the test based on what's on your paper, not on an imagined ideal measurement.

Step-by-Step: Running a Similarity Check From Scratch

This sequence assumes you have a figure and need to establish (or refute) similarity. Follow the order; it minimizes wasted work.

1: Label corresponding parts. Write the letters of the two triangles in matching order, and mark corresponding vertices. If the figure doesn't tell you, infer correspondence from parallel lines or stated angles, then commit to an order. 2: Write the similarity statement you intend to prove. For example, "Prove Ξ”ABC ~ Ξ”DEF." This defines what "correct" means for the rest of the steps. 3: Inventory what you have. Count angles you can calculate or measure, and side lengths you have. Note whether any angle is the included angle between two measured sides. 4: Choose the test with your minimum evidence. Two known angles β†’ AA. Two proportional side ratios plus an included angle β†’ SAS. Three side ratios β†’ SSS. 5: Verify the raw data before the test. Calculate side ratios explicitly to the same precision; identify each included angle; compute the third angle if using AA as a backup. 6: State the test by name in the proof. Write "By AA similarity, Ξ”ABC ~ Ξ”DEF," with the correspondence in the correct order. 7: Check the conclusion for consistency. If you need a side length later, apply the scale factor to the correct pair; if you need an area, square the scale factor.

If a step fails, check the failure mode before repeating the work. Ratios that nearly match but don't β€” 2 and 2.2 β€” usually indicate rounded measurements rather than a genuine similarity failure. Measure again with more precision. If the angles match but the sides looked proportional and aren't, one of the measurements is wrong; use the reliable pair and the reliable angle to recompute.

Worked Calculation: Scale Factor, Missing Side, and Area Change

Take two triangles proven similar by AA with a scale factor of 1.5 from small to large. The small triangle has sides 4, 6, and 8, and the corresponding large triangle's shortest side is 6. Find the remaining large sides and compare areas.

The scale factor confirms the correspondence: 4 Γ— 1.5 = 6. So the remaining large sides are 6 Γ— 1.5 = 9 and 8 Γ— 1.5 = 12. Check with the perimeter: the small perimeter is 4 + 6 + 8 = 18, and the large perimeter is 6 + 9 + 12 = 27. The ratio 27/18 = 1.5, so perimeter scales by the same factor.

Now area. Without the base and height, use the property directly: for similar figures, the area ratio equals the square of the linear scale factor. Here area(large) / area(small) = 1.5Β² = 2.25. If the small triangle's area is 12 square units, the large triangle's area is 12 Γ— 2.25 = 27 square units.

This difference between linear scaling and area scaling is where students lose marks. Two triangles similar with scale factor 3 differ by 3 in side length, 3 in perimeter, and 9 in area. State the factor you're applying, and check whether the answer is a length or an area before finishing. The same logic applies to volume in 3D β€” similarity of cubes scales volume by the cube of the factor.

Three Errors That Wreck Similarity Proofs

The first error is confusing similarity with congruence. A problem says "prove the triangles are similar," and a student proves all three sides equal, which is stronger but sometimes not what the evidence supports. Congruence requires equal lengths; similarity only requires proportional ones.

The second is using an angle that isn't included between the measured sides and calling it SAS. Two side ratios and an angle sitting elsewhere in the triangle tell you almost nothing β€” the SSA arrangement is non-rigid. The fix is to identify the shared vertex of the two measured sides and take the angle at that vertex.

The third is misordering the letters in the similarity statement. "Ξ”ABC ~ Ξ”EDF" asserts angle A corresponds to angle E β€” different from "Ξ”ABC ~ Ξ”DEF," which pairs A with D. When corresponding parts are inconsistent, the scale factor written from the statement won't match the figure.

A fourth error shows up in applied settings: applying similarity to shapes that aren't triangles without converting the method. Two similar quadrilaterals cannot be proved similar by AA, because fixing two angles in a quadrilateral leaves two other angles plus side proportions unspecified. The tests in this article hold for triangles only.

Where Similarity Earns Its Keep in the Real World

Cartography runs on similarity: a map is a scaled copy of terrain, and distances on the map convert to real distances through one scale factor. If a 1:50,000 topographic map shows two points 3 centimeters apart, the real separation is 1.5 kilometers β€” a direct 3 Γ— 50,000 = 150,000 cm computation.

Surveying uses similar triangles to measure distances it can't reach. A surveyor measures the horizontal angle between two points and one distance, then computes the rest via the proportional relationship. The classic mirror method for measuring a tree's height depends on two similar triangles formed by the observer's eye, the mirror, and the tree's top.

Architecture and industrial design use scale models that are physically smaller but proportionally faithful. A model at 1:100 proportion means each interior wall in the model predicts the real wall's position to within build tolerance β€” provided the model is genuinely similar, not just visually close.

Computer graphics embed similarity when models are scaled between scenes. Shadow estimation, camera calibration, and texture placement assume the triangles in the mesh preserve shape under transformation. If you're learning similarity for an exam, you are ultimately learning a vocabulary of proportional reasoning that recurs through every one of these applications.

FAQ

Does AA similarity require all three angles to be measured?

No β€” two measured angles are enough, because the third angle in any planar triangle is forced by the 180-degree sum. If angle A = 35 degrees and angle B = 75 degrees, angle C = 70 degrees by arithmetic. The third-angle check is useful for showing work in a proof, but measurement is not required under AA.

Can two triangles with proportional sides fail to be similar?

Yes, if only two side ratios match β€” the third pair is the test. Suppose two triangles have side ratios 1.5 in two corresponding pairs but 1.7 in the third; SSS fails and the triangles are not similar. The classic error is stopping after two ratios. Every pair must agree under SSS.

Is similarity the same as congruence?

No. Similarity allows size to differ by any scale factor; congruence requires a scale factor of exactly 1. Every pair of congruent triangles is similar, but similar triangles are rarely congruent. State which one you're proving before writing the proof.

How do you tell which angle is the included one?

The included angle is the angle formed by the two measured sides meeting at a shared vertex. If AB and AC are measured, the included angle is at vertex A, not B or C. If you measure AB and BC instead, the included angle moves to vertex B. Mark the shared vertex physically on the figure before applying SAS.

What's the area ratio between two similar triangles with a scale factor of 4?

Sixteen. Area scales by the square of the linear scale factor, so 4Β² = 16. If the smaller triangle's area is 7 square units, the larger triangle's area is 112 square units. Confusing this with the linear factor of 4 produces an area that is four times too small.

Can you use AA on a sphere or in 3D geometry?

Not without modification. On a sphere the angles of a triangle sum to more than 180 degrees, so two matched angles don't force the third β€” check all three. In 3D, similarity extends to similar solids, but the tests differ from the triangle ones.

What if the ratios are close but not exact β€” is it round-off?

Usually. Measured side lengths carry precision limits, so ratios that differ in the last digit β€” 2.00 versus 2.01 β€” typically indicate rounded inputs rather than genuine dissimilarity. Re-measure with tighter tolerances before declaring the triangles dissimilar, particularly in field work where instruments have stated precision.

Does similarity survive rotation, reflection, and translation?

Yes. Similarity is defined up to rotation, reflection, translation, and uniform scaling. Any combination of those transformations preserves shape and angle measures. Non-uniform scaling β€” stretching one axis more than another β€” destroys similarity, which is why resizing a photograph with the aspect ratio locked preserves shapes and resizing without it does not.