Thin-Film Interference & Newton's Rings

A soap bubble is colourless soapy water, yet stretch it into a film and it blazes with shifting bands of gold, magenta, and green; a drop of oil on a wet road does the same. No pigment is involved — the colour is pure interference. Light reflecting off the top of the film meets light reflecting off the bottom, the two beams differ in the distance they travelled by roughly twice the film thickness, and wherever that path difference lands on a bright condition for a particular colour, that colour is reinforced and the film flashes it back. Tilt the bubble or let it drain thinner and the winning wavelength changes, so the colours flow. This is thin-film interference, and it is the same two-source interference of the previous articles with one crucial new ingredient. When light reflects at a surface going into a denser medium — from air into the film — it suffers a sudden half-wavelength phase flip, an extra $\lambda/2$ that reflection into a rarer medium does not. That single half-wave is the whole subtlety of the topic: it silently swaps the naive bright and dark conditions, so that a soap film in air is bright in reflection when $2\mu t = (m+\tfrac12)\lambda$ and dark when $2\mu t = m\lambda$ . Forget the extra $\lambda/2$ and every answer comes out inverted. The same physics, in a controlled geometry, gives Newton's rings: rest a gently curved lens on a flat glass plate and the thin wedge of trapped air produces concentric rings, dark at the centre, at radii $r = \sqrt{m\lambda R}$ — a pattern precise enough to measure a lens's radius of curvature or a wavelength of light. And run the argument in reverse, deliberately arranging cancellation, and you get the anti-reflection coating on every camera lens and pair of glasses: a quarter-wave layer whose two reflections cancel the glare. This article derives the conditions with the phase flip built in, and locks them to numbers you can check.

Key takeaways

Light reflected from the top and bottom of a thin film interferes; their path difference is set by the film's optical thickness $\mu t$ .
Reflection into a denser medium adds an extra $\lambda/2$ (a $\pi$ phase flip); reflection into a rarer medium does not — this is the whole subtlety.
For a soap film in air (only the top reflection flips), reflected light is bright when $2\mu t = (m+\tfrac12)\lambda$ and dark when $2\mu t = m\lambda$ .
Use the wavelength inside the film via the optical thickness $\mu t$ , never the vacuum $\lambda$ with the geometric thickness.
Newton's rings (a lens on a flat plate, an air wedge $t = r^2/2R$ ): reflected rings are dark at $r = \sqrt{m\lambda R}$ (centre dark) and bright at $r = \sqrt{(m-\tfrac12)\lambda R}$ .
Anti-reflection coatings invert the logic: both surfaces flip, so a quarter-wave layer $t = \lambda/4n$ makes the reflections cancel.
Key idea: thin films interfere their top and bottom reflections over an optical path $2\mu t$ , and the extra $\lambda/2$ picked up on reflection into a denser medium sets which thicknesses are bright and which are dark.

The core insight

Two reflections from the faces of a thin film interfere over an optical path $2\mu t$ , and because reflection into a denser medium adds an extra $\lambda/2$ , a soap film in air is bright when $2\mu t = (m+\tfrac12)\lambda$ and dark when $2\mu t = m\lambda$ — the same reasoning giving Newton's dark rings at $r = \sqrt{m\lambda R}$ and the quarter-wave anti-reflection coating $t = \lambda/4n$ .

Where this lives today

Thin-film interference is quietly everywhere. Every quality camera lens, telescope, microscope, and pair of spectacles carries an anti-reflection coating designed by equation (3), and high-end optics stack many quarter-wave layers to kill reflection across the whole visible band — the same multilayer trick that makes a dielectric mirror reflect one colour almost perfectly. The shimmering colours of a butterfly's wing, a peacock's feather, and an opal are structural colours from natural thin films and layered nanostructures, not pigments. Engineers gauge the thickness of oxide layers on silicon chips and of coatings on glass by reading their interference colour to nanometre precision, and the Fabry–Pérot filters in telecom and spectroscopy are thin-film interference pushed to many beams. Newton's rings themselves remain a standard optical-shop test: press a lens against a reference flat and count the rings to check the curvature to a fraction of a wavelength. The soap bubble that opened this article is the same physics that certifies a telescope mirror.

The physical picture

Shine light on a transparent film — soap, oil, or a coating a few hundred nanometres thick. Part of the beam reflects off the top surface; the rest enters the film, crosses it, reflects off the bottom surface, crosses back, and emerges nearly parallel to the first reflection. These two reflected beams come from one incoming wave, so they are coherent, and they interfere. The second beam travelled an extra distance of about $2t$ through the film (near-normal incidence), but distance inside the film is measured in the film's own shorter wavelength, so the meaningful quantity is the optical path $2\mu t$ , where $\mu$ is the film's refractive index. That optical path difference, run through the bright and dark conditions, decides what the film does to each colour. Now the new ingredient. A wave reflecting off a boundary where the medium ahead is optically denser (higher index) is flipped upside down — its phase jumps by $\pi$ , equivalent to adding half a wavelength to its path. Reflecting off a boundary into a rarer medium gives no such flip. For a soap film in air the top reflection is air-into-film (into denser: it flips) while the bottom reflection is film-into-air (into rarer: no flip). So one of the two beams carries an extra $\lambda/2$ that the other does not, and that half-wave must be folded into the interference bookkeeping. This is why the conditions below look "backwards" compared with two beams that simply differ by $2\mu t$ : the lone flip has swapped bright and dark.

Figure

Thin-film interference — the top reflection flips by half a wavelength on entering the denser film while the bottom reflection does not, and the two interfere over the optical path 2 mu t

The reflected-light conditions

Combine the two effects — the optical path $2\mu t$ and the single $\lambda/2$ flip at the top — and the interference conditions for the reflected light drop out. The two reflected beams reinforce (a bright film in reflection) when their total effective path difference is a whole number of wavelengths. Because the flip already contributes half a wavelength, the geometric part must supply the missing half, giving

2\mu t = \left(m + \tfrac12\right)\lambda, \qquad m = 0, 1, 2, \dots \qquad (1)

where: $\mu$ is the film's refractive index, $t$ its thickness, $\lambda$ the vacuum wavelength, and $m$ the order. The physical reading is that the thinnest bright film ( $m = 0$ ) has $2\mu t = \lambda/2$ , i.e. optical thickness a quarter of a wavelength. The film is dark in reflection when the two beams oppose, which — again with the flip folded in — happens at

2\mu t = m\lambda, \qquad m = 0, 1, 2, \dots \qquad (2)

The $m = 0$ dark case is $t = 0$ : a film far thinner than a wavelength looks black in reflection, because the tiny geometric path leaves only the bare $\lambda/2$ flip, which cancels the two beams. This is exactly the black patch that spreads across the top of a soap film just before it bursts. Put in soap numbers, $\mu = 1.33$ and $\lambda = 600\ \mathrm{nm}$ . The brightest (thinnest) reflecting film from equation (1) with $m = 0$ :

t = \frac{\lambda}{4\mu} = \frac{600\ \mathrm{nm}}{4(1.33)} = 112.8\ \mathrm{nm} \;(\approx 113\ \mathrm{nm}),

and the first dark film from equation (2) with $m = 1$ :

t = \frac{\lambda}{2\mu} = \frac{600\ \mathrm{nm}}{2(1.33)} = 225.6\ \mathrm{nm}.

So as a soap film drains from thick to thin, it runs through a sequence of bright and dark thicknesses for each colour — and because the winning thickness differs for red, green, and blue, you see the flowing bands of colour.

Anti-reflection coatings: the logic reversed

Deliberately arrange for the reflections to cancel and you suppress glare. Coat glass ( $n_{\text{glass}} = 1.5$ ) with a thin film of magnesium fluoride ( $n = 1.38$ ), chosen because its index lies between air and glass. Now both reflections go into a denser medium — air into MgF $_2$ (denser) and MgF $_2$ into glass (denser still) — so both flip, and the two $\lambda/2$ flips cancel, leaving no net extra half-wave. With the flips gone, cancellation of the reflected light (minimum reflection) needs the geometric optical path alone to be a half-wavelength, $2nt = (m+\tfrac12)\lambda$ ; the thinnest such quarter-wave coating is

t = \frac{\lambda}{4n} = \frac{550\ \mathrm{nm}}{4(1.38)} = 99.6\ \mathrm{nm} \qquad (3)

where: $n$ is the coating index and $\lambda = 550\ \mathrm{nm}$ is chosen near the eye's peak sensitivity. The physical reading is that a layer a quarter-wavelength thick makes the two reflections a half-wavelength apart, so they interfere destructively and the reflection vanishes — the faint purple sheen on a good camera lens is the residual of the wavelengths this thickness does not quite kill. Note the sign flip in the argument: because both surfaces now flip, the quarter-wave layer cancels the reflection, whereas for a soap film (one flip) the same quarter-wave optical thickness made it brightest.

Newton's rings

Give the air film a controlled shape and you can measure with it. Rest a plano-convex lens, curved side down, on a flat glass plate: between them lies a wedge of air, zero thickness at the point of contact and growing outward. By the sagitta geometry of a shallow curve, the air-gap thickness at radius $r$ from the contact point is

t = \frac{r^2}{2R} \qquad (4)

where: $R$ is the lens's radius of curvature (this geometric result is quoted, not derived here). The two reflections are now off the top and bottom of the air film ( $\mu = 1$ ): the top reflection glass-into-air (rarer, no flip) and the bottom reflection air-into-glass (denser, flips). One flip again, so the dark rings in reflection occur where $2t = m\lambda$ . Substituting equation (4):

r = \sqrt{m\lambda R}, \qquad m = 0, 1, 2, \dots \qquad (5)

where: $r$ is the radius of the $m$ -th dark ring. At the centre, $r = 0$ gives $m = 0$ — a dark spot, because the two glass surfaces touch, the geometric path is zero, and only the bare $\lambda/2$ flip remains to cancel the beams. The bright rings sit between, at $r = \sqrt{(m-\tfrac12)\lambda R}$ . For $R = 1.0\ \mathrm{m}$ and $\lambda = 600\ \mathrm{nm}$ :

r_1 = \sqrt{\lambda R} = \sqrt{(600\times10^{-9})(1.0)} = 0.775\ \mathrm{mm}, \quad r_2 = \sqrt{2\lambda R} = 1.095\ \mathrm{mm},

and the first bright ring at $r = \sqrt{0.5\,\lambda R} = 0.548\ \mathrm{mm}$ . The dark radii grow as $\sqrt{m}$ , so the rings crowd closer together farther out — the signature look of the pattern. Measuring the ring radii and knowing $\lambda$ gives $R$ ; knowing $R$ gives $\lambda$ . This is the quantitative use that made the rings a metrology tool for two centuries.

Figure

Newton's rings — a lens on a flat plate traps an air wedge of thickness r-squared over 2R, giving reflected rings that are dark at the centre with radii root m lambda R

Derivation AThe reflected bright and dark conditions with the extra half-wavelength

The goal is to obtain the reflected-light conditions for a soap film in air — bright at $2\mu t = (m+\tfrac12)\lambda$ and dark at $2\mu t = m\lambda$ — by tracking the two reflections' phases, including the half-wave flip on reflection into the denser film. The whole point is to see why the conditions look inverted, so we build the total phase difference piece by piece; the mathematics is arithmetic of half-wavelengths, no calculus.

Step 1. List the two reflected beams. Beam 1 reflects off the top surface, air into film: the film is denser, so this reflection flips, adding $\lambda/2$ to its effective path. Beam 2 enters the film, reflects off the bottom (film into air, rarer, no flip), and returns, having travelled an extra geometric distance $\approx 2t$ inside a medium of index $\mu$ , i.e. an extra optical path $2\mu t$ .

Step 2. Write the total effective path difference between the two beams. Beam 2 leads by the optical path $2\mu t$ ; beam 1 leads by the half-wave from its flip. Net effective difference $= 2\mu t + \tfrac{\lambda}{2}$ (taking beam 2's longer travel and beam 1's flip together in the reflected wave).

Step 3. Apply the interference conditions to the effective difference. Constructive (bright): $2\mu t + \tfrac{\lambda}{2} = m'\lambda$ for integer $m'$ , which rearranges to $2\mu t = (m' - \tfrac12)\lambda = (m + \tfrac12)\lambda$ on relabelling $m = m' - 1 = 0, 1, 2, \dots$ — equation (1). Destructive (dark): $2\mu t + \tfrac{\lambda}{2} = (m + \tfrac12)\lambda$ , giving $2\mu t = m\lambda$ — equation (2).

Step 4. Read the landmark thicknesses for $\mu = 1.33$ , $\lambda = 600\ \mathrm{nm}$ . Thinnest bright ( $m = 0$ ): $t = \lambda/4\mu = 112.8\ \mathrm{nm}$ . First dark beyond black ( $m = 1$ ): $t = \lambda/2\mu = 225.6\ \mathrm{nm}$ . The $m = 0$ dark case is $t = 0$ , the black film — the tell-tale that the lone flip is doing the work. Verified: $t_{\text{bright}} = \lambda/4\mu = 112.8\ \mathrm{nm}$ (≈113 nm), $t_{\text{dark}} = \lambda/2\mu = 225.6\ \mathrm{nm}$ — recomputed in verify_hs2-thin-film-interference.py.

Go deeper · Derivation B · 1 minNewton's dark-ring radius r equals root m lambda R

Derivation BNewton's dark-ring radius r equals root m lambda R

The goal is to derive the dark-ring radius $r = \sqrt{m\lambda R}$ for Newton's rings by combining the air-wedge geometry with the dark-interference condition, and to confirm the centre is dark. We quote only the sagitta relation $t = r^2/2R$ (a standard result for a shallow spherical cap); the rest is the same half-wave bookkeeping as the soap film, applied to an air film. No calculus.

Step 1. Get the film thickness at radius $r$ . A sphere of radius $R$ resting on a plane rises a height $t = r^2/2R$ above the contact point at horizontal distance $r$ (the sagitta approximation for $r \ll R$ ). This is the air-gap thickness that varies across the pattern.

Step 2. Set the reflection phases. The air film has the glass above and glass below. Top reflection is glass into air (rarer, no flip); bottom reflection is air into glass (denser, flips, adding $\lambda/2$ ). One flip, exactly as for the soap film, and the film index here is $\mu = 1$ (air).

Step 3. Impose the dark condition. With one flip, the two beams cancel (dark) when the geometric optical path is a whole number of wavelengths: $2\mu t = 2t = m\lambda$ , so $t = m\lambda/2$ . Substitute the geometry $t = r^2/2R$ : $r^2/2R = m\lambda/2$ , hence $r^2 = m\lambda R$ and $r = \sqrt{m\lambda R}$ — equation (5). The centre ( $r = 0$ ) gives $m = 0$ : a dark spot, since only the $\lambda/2$ flip survives there.

Step 4. Evaluate for $R = 1.0\ \mathrm{m}$ , $\lambda = 600\ \mathrm{nm}$ . Dark rings: $r_1 = \sqrt{\lambda R} = 0.775\ \mathrm{mm}$ , $r_2 = \sqrt{2\lambda R} = 1.095\ \mathrm{mm}$ (so $r_2/r_1 = \sqrt2 = 1.414$ ). Bright rings sit at $r = \sqrt{(m-\tfrac12)\lambda R}$ , giving a first bright ring at $\sqrt{0.5\,\lambda R} = 0.548\ \mathrm{mm}$ . Verified: $r_1 = \sqrt{\lambda R} = 0.775\ \mathrm{mm}$ , $r_2 = \sqrt{2\lambda R} = 1.095\ \mathrm{mm}$ ( $r_2/r_1 = \sqrt2$ ), bright $r_1 = \sqrt{0.5\lambda R} = 0.548\ \mathrm{mm}$ — recomputed in verify_hs2-thin-film-interference.py.

Limiting cases

case	result	meaning
$t \to 0$ (soap)	dark (black film)	only the $\lambda/2$ flip remains; the beams cancel
$2\mu t = \lambda/2$ ( $t = \lambda/4\mu$ )	brightest reflection	thinnest bright film, $m = 0$
$2\mu t = m\lambda$	dark reflection	destructive with the flip folded in
Newton's centre, $r = 0$	dark spot	contact point, one flip, zero path
dark ring radii	$r \propto \sqrt{m}$	rings crowd together farther out
AR coating (both flip)	quarter-wave $t = \lambda/4n$ cancels	two flips cancel, so a quarter-wave suppresses reflection
use vacuum $\lambda$ and $t$	wrong answer	must use the optical thickness $\mu t$

Common trapDo not forget the extra half-wavelength, and use the optical thickness

Two mistakes wreck thin-film problems. First, and most costly, is forgetting the extra $\lambda/2$ picked up on reflection into a denser medium. That single half-wave swaps every bright and dark condition, so a soap film in air is bright at $2\mu t = (m+\tfrac12)\lambda$ , not at $m\lambda$ . Always ask, at each surface, "is the medium ahead denser?" — if yes, that reflection flips. Second, the path inside the film is an optical path: use $2\mu t$ , not $2t$ , because the wavelength inside is $\lambda/\mu$ . Writing the condition with the vacuum wavelength and the geometric thickness drops the factor $\mu$ . Finally, remember that Newton's centre is dark in reflection (not bright), and that anti-reflection coatings invert the soap-film logic because both their surfaces flip.

Worked examples

Board level — the thinnest bright and first dark soap film (ISC and NEET pattern)

A soap film of refractive index $1.33$ is viewed in reflected light of wavelength $600\ \mathrm{nm}$ at near-normal incidence. Find (a) the smallest thickness for which the film appears brightest, and (b) the smallest non-zero thickness for which it appears dark.

Step 1. Brightest film: use equation (1), $2\mu t = (m+\tfrac12)\lambda$ , with the smallest $m = 0$ : $2\mu t = \tfrac12\lambda$ , so $t = \lambda/4\mu = (600\times10^{-9})/(4\times1.33) = \boxed{112.8\ \mathrm{nm}}$ (about $113\ \mathrm{nm}$ ).

Step 2. First dark film: use equation (2), $2\mu t = m\lambda$ . The $m = 0$ case is $t = 0$ (the black film), so the smallest non-zero dark thickness is $m = 1$ : $t = \lambda/2\mu = (600\times10^{-9})/(2\times1.33) = \boxed{225.6\ \mathrm{nm}}$ .

Step 3. Notice the pattern: the dark thickness is exactly twice the bright one ( $225.6 = 2\times112.8\ \mathrm{nm}$ ), because $\lambda/2\mu = 2\times(\lambda/4\mu)$ — bright and dark thicknesses alternate in steps of $\lambda/4\mu$ .

Step 4. Sense-check the flip: had we forgotten the extra $\lambda/2$ , we would have called $t = 0$ bright, but a vanishingly thin soap film is observed to be black in reflection. The flip is what makes the physics match the bubble.

Competitive level — quarter-wave anti-reflection coating (JEE and AP-2 pattern)

A camera lens of glass ( $n_{\text{glass}} = 1.5$ ) is to be coated with magnesium fluoride ( $n = 1.38$ ) so that reflection of $550\ \mathrm{nm}$ light is minimised. Find the thinnest coating, and explain why the same quarter-wave thickness makes a soap film brightest but this coating darkest in reflection.

Step 1. Both reflections flip here: air into MgF $_2$ (denser) flips, and MgF $_2$ into glass (denser, since $1.5 > 1.38$ ) flips too. Two flips cancel, so there is no net extra $\lambda/2$ .

Step 2. With no net flip, minimum reflection (destructive) needs the geometric optical path to be a half-wavelength: $2nt = (m+\tfrac12)\lambda$ . Thinnest is $m = 0$ : $t = \lambda/4n = (550\times10^{-9})/(4\times1.38) = \boxed{99.6\ \mathrm{nm}}$ .

Step 3. Why the same quarter-wave thickness gives opposite results: for the soap film (one flip) the quarter-wave optical thickness put the beams a whole wavelength apart after the flip — constructive, brightest. For the AR coating (two flips, which cancel) the same quarter-wave puts the beams a half wavelength apart — destructive, darkest. The number of flips flips the outcome.

Step 4. The residual: a single quarter-wave layer perfectly kills only $550\ \mathrm{nm}$ ; red and blue are suppressed less, which is why a coated lens shows a faint purple reflection — the leftover of the wavelengths this thickness does not exactly cancel. Timing note: about 75 seconds. The trap is the flip count — before writing any condition, tally how many of the two reflections go into a denser medium; an even number cancels, an odd number leaves an extra $\lambda/2$ .

Exam traps

Count the flips first. At each surface, ask whether the medium ahead is denser. An odd number of denser-medium reflections leaves an extra $\lambda/2$ ; an even number cancels. This single tally decides whether your conditions are $2\mu t = (m+\tfrac12)\lambda$ or $2\mu t = m\lambda$ for bright.

Optical thickness, not geometric. The path inside the film is $2\mu t$ , because the wavelength there is $\lambda/\mu$ . Do not use the vacuum $\lambda$ with the bare thickness $t$ — you would lose the factor $\mu$ .

Newton's centre is dark in reflection. The contact point has zero air gap, so only the lone flip acts and the beams cancel: a dark spot. Writing "bright centre" inverts the whole ring count. (In transmitted light the pattern is complementary — bright centre — but exam questions almost always mean reflection.)

Dark ring radii go as $\sqrt{m}$ . From $r = \sqrt{m\lambda R}$ , the rings are not equally spaced; they bunch up outward. Do not treat successive ring radii as an arithmetic sequence.

AR coatings invert the condition. Because both coating surfaces flip, a quarter-wave layer cancels reflection, the opposite of the single-flip soap film. Match the condition to the number of flips, not to a memorised formula.

Board scope. Thin-film interference and Newton's rings are assessed on CBSE, ICSE, JEE, NEET, and AP Physics 2; they are not in the IB 2023 syllabus or A-Level (CIE 9702), so do not tell an IB or A-Level student to revise them. NCERT treats film colours only qualitatively — the quantitative Newton's-rings work ( $r = \sqrt{m\lambda R}$ ) is ISC, JEE, and NEET depth; AP-2 firmly carries thin-film interference including the $\lambda/2$ reflection shift.

Origins · 1 minIsaac Newton and how this was discovered

Isaac Newton

1643–1727 · Woolsthorpe & Cambridge

Isaac Newton is the towering figure of classical physics — the laws of motion, universal gravitation, and a share of the calculus — and in optics he split white light into its spectrum and built the first reflecting telescope. The pattern that bears his name, Newton's rings, he studied around the 1670s and catalogued meticulously in his 1704 Opticks, measuring the ring radii and the corresponding air-gap thicknesses with remarkable care. The irony is deep: Newton favoured a corpuscular (particle) theory of light and resisted the wave view, yet the rings he mapped are a textbook wave-interference effect, and it took Thomas Young, more than a century later, to explain them by interference and the half-wave phase change on reflection. Newton's data were so good that they later served as evidence against his own theory — a reminder that careful measurement outlives the model that motivated it.

History

1665

— Robert Hooke reports the colours of thin plates of mica and soap in his Micrographia.

1675

— Isaac Newton studies the rings formed between a convex lens and a flat plate and measures their radii.

1704

— Newton publishes Opticks, tabulating the ring thicknesses despite favouring a corpuscular theory of light.

1801–1804

— Thomas Young explains the colours and Newton's rings by interference and announces the half-wavelength phase change on reflection.

1935

— Alexander Smakula patents the anti-reflection coating, applying thin-film interference to suppress glare on lenses.

Going further

Texts. NCERT Physics Class 12, Chapter 10 ("Wave Optics"), the qualitative treatment of thin-film colours · H. C. Verma, Concepts of Physics Vol. 1, "Light Waves", thin films · Irodov / ISC references for the quantitative Newton's-rings analysis.

Provenance. Standard results. Recompute: soap brightest $t = \lambda/4\mu = 600/(4\times1.33) = 112.8\ \mathrm{nm}$ (≈113 nm); first dark $t = \lambda/2\mu = 225.6\ \mathrm{nm}$ ; MgF $_2$ AR coating $t = \lambda/4n = 550/(4\times1.38) = 99.6\ \mathrm{nm}$ ; Newton dark $r_1 = \sqrt{\lambda R} = \sqrt{(600\ \mathrm{nm})(1\ \mathrm{m})} = 0.775\ \mathrm{mm}$ , $r_2 = \sqrt{2\lambda R} = 1.095\ \mathrm{mm}$ (ratio $\sqrt2$ ), bright $r_1 = \sqrt{0.5\lambda R} = 0.548\ \mathrm{mm}$ .

Prerequisites. Coherence & Interference: Path Difference, Phase & Intensity — the bright and dark conditions this article adds a reflection flip to. Leads to. the university seams Thin-Film Interference and Newton's Rings (go-deeper only, never an exam dependency here), which handle oblique incidence and multiple beams.

Why does a very thin soap film look black just before it bursts?

Because when the film gets much thinner than a wavelength, the extra geometric path $2\mu t$ shrinks toward zero, and the only thing left is the half-wavelength flip that the top reflection picked up. With one beam flipped by $\lambda/2$ and the other essentially unshifted, the two reflected waves are exactly out of step and cancel — no light is reflected, so the film looks black. This is the $m = 0$ case of the dark condition $2\mu t = m\lambda$ , which gives $t = 0$ . You can watch it happen: as a vertical soap film drains, the top thins first and a black band spreads downward moments before the film pops. The blackness is a direct visual proof of the reflection phase flip.

What exactly is the "extra λ/2" and when do I add it?

It is a half-wavelength jump in path that a wave suffers when it reflects at a surface with a denser (higher-index) medium on the far side — equivalently a $\pi$ phase flip, the wave turning upside down. Reflection into a rarer medium gives no such jump. You fold it in by asking, at each of the two film surfaces, whether the medium the light is about to enter is denser: if so, that reflection adds $\lambda/2$ . For a soap film in air only the top surface qualifies (air into film), so there is one flip and the conditions are $2\mu t = (m+\tfrac12)\lambda$ for bright. For a coating between air and glass, both surfaces can flip, the two cancel, and the conditions revert. Count the flips and the sign takes care of itself.

Why must I use 2μt and not just 2t for the path difference?

Because the second beam travels its extra distance inside the film, where the wavelength is shorter than in vacuum by the factor $\mu$ (the frequency is fixed, the speed is $c/\mu$ , so $\lambda_{\text{film}} = \lambda/\mu$ ). Interference counts how many wavelengths fit into the extra path, and since each wavelength inside the film is shorter, the same geometric distance $2t$ contains $\mu$ times as many of them. The quantity that measures "number of wavelengths of path" is therefore the optical path $2\mu t$ , not the geometric $2t$ . Using $2t$ with the vacuum wavelength drops the factor $\mu$ and gives thicknesses wrong by that factor — a common and easily avoided slip.

Why is the centre of Newton's rings dark and not bright?

At the exact centre the lens touches the plate, so the trapped air film has zero thickness and the extra geometric path is zero. But the bottom reflection (air into glass, a denser medium) still flips by $\lambda/2$ , while the top reflection (glass into air) does not. With one beam flipped and no compensating path difference, the two reflected waves are half a wavelength out of step and cancel — a dark spot. This is the $m = 0$ dark ring. It is a nice consistency check on the whole analysis: if you ever compute a bright centre in reflection, you have dropped the phase flip. (Viewed in transmitted light the centre is bright, because the transmitted beams have the complementary phase relationship.)

How are Newton's rings used to measure a wavelength or a radius of curvature?

Rearrange the dark-ring formula $r^2 = m\lambda R$ . If you know the lens's radius of curvature $R$ , measure the radius $r_m$ of the $m$ -th dark ring and solve $\lambda = r_m^2/(mR)$ to get the wavelength. If instead you know $\lambda$ (from a laser, say), measure the rings and solve $R = r_m^2/(m\lambda)$ to get the curvature. In practice you measure several rings and use the difference $r_m^2 - r_n^2 = (m-n)\lambda R$ , which cancels errors from not knowing exactly where the centre is and from any dust at the contact point. Because ring radii of a fraction of a millimetre can be read to microns under a travelling microscope, the method is precise enough to certify optical components.

Why do anti-reflection coatings leave a faint purple or magenta sheen?

Because a single quarter-wave layer is tuned to cancel just one wavelength perfectly — usually green, near $550\ \mathrm{nm}$ , where the eye is most sensitive. At that wavelength the two reflections are half a wavelength apart and cancel completely. But red and blue light have different wavelengths, so for them the layer's optical thickness is not quite a quarter-wave, the cancellation is incomplete, and a little red and blue are still reflected. Red plus blue reads to the eye as purple or magenta, which is the residual glare you see on a coated lens. Multi-layer coatings stack several films to flatten the cancellation across the whole visible range, which is why premium optics reflect an even fainter, more neutral tint.