That numbers exist is not in doubt, and a page claiming otherwise would be wrong. Five of the positions read on 26 September 2026 print millimetres for the width of a front, and three publish them as a ladder of named rungs: one puts its band in a table whose header says the frame is measured from hinge to hinge, one sells six rungs from Junior to Extra Large and prints forty-six millimetre figures across three tables before routing its face measurement into the two quantities that are not width, and one names four rungs after a width measured inside the front. The remaining two print a number without a ladder: one a target of 136 mm with a two-millimetre window around it, one a width measured centre to centre at its own hinge screws and tabulated model by model. What nobody in the set does is the step that would make any of those bands usable. Three pages each carry a rung called Medium, at 129 to 132 mm, 129 to 134 mm and 126 to 130 mm: face to face that is a disagreement of 3 mm (0.12 in) at the bottom of the rung and 4 mm (0.16 in) at the top, or one and a half to two times the two-millimetre tolerance that one of those pages prints as its own advice. Read onto a single measurement point — which is arithmetic on a published median, not a preference — the same three rungs stand 13 to 20 mm apart, six and a half to ten times that tolerance. The middle of this page is the subtraction that makes the conversion, and the last column of the first table is what it leaves you holding.
What one stamped size is worth, measured four ways
Rows are stamped sizes — the two numbers a frame prints, lens width then bridge width — all inside the published adult bands of 40 to 60 mm for a lens and 14 to 24 mm (0.55 to 0.94 in) for a bridge. Columns are quantities a front has whether or not anybody measures it. Column two is lens plus bridge, the sum this site's page about the stamped numbers calls the frame PD. Column three is two lenses plus the bridge; that is how this site reads a published hinge-to-hinge ladder. Columns four and five add a pair end-piece allowance to column three: 13 mm (0.51 in) is the metal median and 20 mm (0.79 in) the acetate median, both published by this site in August 2026 from 3,942 retailer frame records, and printed on no frame anywhere. Column six subtracts a 136 mm (5.35 in) cheekbone line, the middle case this site works on its heart-shaped-face page; the published adult cheekbone band runs 128.2 to 148.8 mm (5.05 to 5.86 in), so every figure in that column moves by +7.8 mm (0.31 in) at the band's lower edge and −12.8 mm (0.50 in) at its upper edge. The rows are arithmetic on two stamped digits and one published assumption, not stock items this site measured.
| As stamped | Lens centres apart | Hinge to hinge | Front at a metal rim | Front at an acetate rim | Against a 136 mm cheekbone line |
|---|---|---|---|---|---|
| 48□18 | 66 mm (2.60 in) | 114 mm (4.49 in) | 127 mm (5.00 in) | 134 mm (5.28 in) | 9 mm short, metal · 2 mm short, acetate |
| 50□17 | 67 mm (2.64 in) | 117 mm (4.61 in) | 130 mm (5.12 in) | 137 mm (5.39 in) | 6 mm short, metal · 1 mm over, acetate |
| 52□16 | 68 mm (2.68 in) | 120 mm (4.72 in) | 133 mm (5.24 in) | 140 mm (5.51 in) | 3 mm short, metal · 4 mm over, acetate |
| 52□18 | 70 mm (2.76 in) | 122 mm (4.80 in) | 135 mm (5.31 in) | 142 mm (5.59 in) | 1 mm short, metal · 6 mm over, acetate |
| 54□18 | 72 mm (2.83 in) | 126 mm (4.96 in) | 139 mm (5.47 in) | 146 mm (5.75 in) | 3 mm over, metal · 10 mm over, acetate |
| 56□16 | 72 mm (2.83 in) | 128 mm (5.04 in) | 141 mm (5.55 in) | 148 mm (5.83 in) | 5 mm over, metal · 12 mm over, acetate |
No account · No ad pixels · No photo upload
Detection runs on-device with MediaPipe. Nothing is uploaded — no account, no email, no storage.
Which width each page is holding, in its own words
Take the two sentences that appear on one page, because that page removes any need to compare publishers. Its subject is exactly this quantity, and it defines it twice. First: “This is the overall width from left to right outer edges of the frame itself.” Later, describing its own diagram, it says the caliper “touches with two ends of the hinge behind.” Those are two line segments on the same object. The first is column four or five of the table above — the front, end piece to end piece. The second is column three, the hinge span, and the diagram's own page numbers its figures so the two instructions sit a paragraph apart. Between them lie both end pieces, and the published medians put 13 to 20 mm (0.51 to 0.79 in) into that space.
The same page then gives a number and a tolerance: “The frame width doesn't have to be exact — if the width is within two millimeters in either direction, it should still fit comfortably. So for a target frame width of 136, if you set the range to 134-138, the glasses will all fit you.” Read that 136 against its own two definitions and it is either a front of 136 mm or a hinge span of 136 mm, which is a front of 149 to 156 mm (5.87 to 6.14 in) on the published medians. The gap between those two readings is six and a half to ten times the tolerance the same sentence asks you to hold it to. Nobody is being careless here more than everyone is: on the ten pages read, the width word also carries a definition measured centre to centre at the temple hinge screws, one measured from hinge to hinge, one measured inside the front, and one that is not measured on the frame at all — “Frame width is the most straightforward aspect and is determined by measuring the distance between your temples,” which is a line on your head wearing the name of a line on the object.
So the question to ask of any published band is not “is this number right” but “which of the five is this”. That is a genuinely new question in this family, and it has a boundary: it is not a claim that nobody distinguishes measurement points. This site's own page about bridge width prints a hinge-to-hinge column beside a front column in the same table, and the twenty millimetres between them are in the table note with a twelve-millimetre sensitivity. What has not been done anywhere, here included until the next table, is to take a published band from a page that does not say which point it uses and move it. Every point named above is measured by at least one page except the distance from an inside edge to an outside edge, and that one is named as a quantity and then left unconverted, because nothing in the set prints it.
Three ladders, one word for the middle rung
The bands that exist are not small in number, and they are not wildly different either. That is what makes the disagreement easy to miss. Three published ladders each have a rung called Medium. Their millimetres, as printed, are 129 to 132, 129 to 134 and 126 to 130. Sorted by face value those overlap almost completely: 3 mm separates the lowest floor from the highest floor, 4 mm (0.16 in) the lowest ceiling from the highest, and a shopper who reads three pages and averages them lands near 130 mm and is inside every one of the three. That is the reading the family invites, and it is the one that survives contact with nothing.
The three rungs are not measured along the same line. The ladder that prints 129 to 132 says in its own table header that the frame is measured horizontally from hinge to hinge, so its Medium is column three. The six-rung ladder that prints 129 to 134 states no measurement point anywhere on the page, in either direction. The four-rung ladder that prints 126 to 130 calls its quantity internal and gives no conversion out. Move the one that tells you what it does — the hinge-to-hinge ladder — onto the outside of the front by adding the published pair allowance, and its Medium becomes 142 to 145 mm on a metal rim and 149 to 152 mm (5.87 to 5.98 in) on an acetate one. That is the same rung the second ladder also calls 129 to 134, and the floors of the two now stand 13 to 20 mm apart. One of the two readings has to be wrong about a size that a shopper can wear, and no page in the set is able to say which, because the only ladders that print a measurement point disagree with each other about it.
Rows are the published width ladders read on 26 September 2026, plus this site's own two rows, which are included so the conversion is applied to the house numbers as well. Column one is the rung each page sells as Medium, in the units and digits it prints. Column two is what the page itself says the measurement runs along; where a page says nothing, the cell says so. Column three adds the pair end-piece allowance published by this site in August 2026 — 13 mm metal, 20 mm acetate — where column two is a hinge span, and leaves the figures alone where the page's quantity is already an outside measure. Column four is what the page says a size has to be within reach of; only one of the four external pages prints a number, so that column is mostly words, and a column of mostly words is itself the finding.
| The rung called Medium | Where the page says the line runs | The same rung read as an outside front | What the page says you have to hit |
|---|---|---|---|
| 129–132 mm (5.08–5.20 in) | “from hinge to hinge”, in its own table header | 142–145 mm metal · 149–152 mm acetate | Nothing — the ladder sorts stock, it does not tell you which rung is yours |
| 129–134 mm (5.08–5.28 in) | Not stated on the page, in either direction | Either 129–134 mm as printed, or 142–147 mm (5.59–5.79 in) metal · 149–154 mm (5.87–6.06 in) acetate on the hinge reading — the page cannot be placed on either line | Nothing — the same page's face-width table routes to lens and arm instead |
| 126–130 mm (4.96–5.12 in) | “internal frame width” | Cannot be converted: no page in this set prints inside-edge to outside-edge | Nothing — a small-or-large verdict with no line drawn to it |
| 136 mm as a target, not a band | Two points, both printed on that page: outer edges of the frame, and the two ends of the hinge behind | 136 mm on the first, 149–156 mm (5.87–6.14 in) on the second | “within two millimeters in either direction”, so 134–138 mm |
| 130–145 mm (5.12–5.71 in), this site, bridge-span reading | “Total width, hinge to hinge”, as this site's oval-face table names it | 143–158 mm (5.63–6.22 in) metal · 150–165 mm (5.91–6.50 in) acetate | “within about 3 mm (0.12 in) of your own cheekbone line” |
| 130–145 mm, the same digits on the same site | “Total width, end piece to end piece”, as this site's heart-shaped-face table names it | 130–145 mm — already an outside measure | The same cheekbone line, at a different point |
The subtraction against a line on your head
Column six of the first table is the step the printed ladders never take, and it is a subtraction with two operands, both of which you can have. Take the front from column four or five. Take your own cheekbone line, the widest measurement on your head that a pair of glasses has to clear, taken temple to temple with a tape; for most adults it sits inside the 128.2 to 148.8 mm published band. Subtract the second from the first. A negative answer means the front stops short of the widest point of the face; a positive one means it reaches past it. The 3 mm that this site's own cheekbone rule allows is the half-width of the window on either side of zero in which nothing is wrong. On the 52□16 row a metal rim sits 3 mm short, at the edge of that window, and an acetate rim 4 mm past it — one stamped size, two verdicts, and both of them arguable, which is why every page that lists the digits can honestly say they fit.
Read the 52□18 row again, because it is the worked example the sizing pages use, and it is where the two complaints live at once. On a metal rim its front is 135 mm and it stops 1 mm short of a 136 mm cheekbone line. On an acetate rim its front is 142 mm and it runs 6 mm past the same line on the same face. One number on the temple of the frame, one measurement on your head, and both verdicts. If you have ever read a page that warns you a frame is going to look and feel too big and thought that glasses too wide for my face is exactly the trouble you had with a size you ordered twice, this row is the mechanism: you were not choosing between a wide frame and a narrow one, you were choosing between two objects that print the same two digits and differ by the material of the rim, which is the one term in the arithmetic that neither the stamp nor any published ladder carries.
The other side of the family's advice is the squeeze, and it gets the same treatment. “When your glasses feel like they're pinching the sides of your head, the frame width is likely too narrow,” prints one of the ten, in a section whose symptom headings are all sensations; the page carries ten millimetre figures elsewhere and none attached to that sentence. Note which word does the work there: the pressure is on the sides of the head, at the temples, which are not the arms of the glasses — the same double name is why a buyer searching for glasses pressing on temples gets pages about the hardware and pages about the skin, and why this page keeps the two apart by writing the line segment every time. The oldest position read for this page puts both complaints on opposite ends of the same quantity and still gives no dimensions: a medical treatise on spectacles published in 1850 argues in one sentence that a framework too large and broad loses its steadiness and tires the eyes through the movement of the image, while one too narrow and heavy compresses the temples and aches near the eye. It is the same paragraph, two sentences later, that adds a third failure mode and puts it on a different width: the glasses should be neither too near nor too far from each other, and if that is not attended to, double vision and other anomalies of vision may result.
That third clause is the one worth dragging 176 years forward, because it is about column two, not column four. The distance between the two lenses is the only width in the first table that a frame prints, and it is the only one with a published consequence attached to a mechanism rather than to a feeling. This site's page about the stamped numbers does exactly that with it: lens plus bridge is the frame PD, and subtracting your pupillary distance from it and halving gives the work a laboratory has to do with the lens. The distance between the outer edges of that same front is the width nobody prints, and it has no such consequence; it has only sensations, and a band that means four things. A buyer who confuses the two columns gets the 1850 treatise's third warning from the wrong operand and the first from the wrong one as well.
Why they slide is not a width question
The other half of what this family promises is that slipping and pinching sit on opposite sides of the width, and the eight pages read for that question were opened expecting to find the line where one turns into the other. Every one of the eight carries a cause list, and across all eight the count of millimetre figures is zero — not zero in the slipping section, zero on the page, in a set whose longest member runs 2,957 words and whose next longest 2,376. The causes are parts and adjectives: frames too large, temples not adjusted properly, loose screws, nose pads badly placed, oily skin. “If the bridge is too wide, the glasses will not stay in position.” “Bridge fit: If the bridge is too wide or narrow, it throws off the balance.” “Sometimes glasses slide because they're too big for your face.” “The arms are too loose or too short.” Two of those four are not about the width of a front at all, and the two that use the word wide are about the bridge and about size in general, which the arithmetic above has already separated.
The direction is worse than the absence of numbers. In the longest of the eight, one wearer explains that though it does not seem to make sense, glasses many times slide because they are too tight behind the ear or on the side of the head, and the reply immediately below asks whether too large was meant. Nobody answers it with a quantity, because there is none on the page to answer with. So if your question is the plain one — why do my glasses slide down my nose — the honest answer this round supports is that sliding has at least three separate mechanisms in the published record, one of them the bridge, one of them the arms behind the ear, one of them the pads and skin, and that the width of a front is not established as a cause anywhere in the eight. A residual from the last section cannot tell you which of the three you have. It can tell you whether the front clears your cheekbones, the subject of this page; the other two are settled on the page that starts from the arms and the page that starts from the middle number.
This is also where the subtraction stops being enough, and the reason has to be stated exactly. A width residual is one number with a sign, and a sign is a direction, not a diagnosis: at −6 mm the front is short of the face and the arms have to run further back to reach past it, a plausible path from a narrow front to a slipping pair of glasses; at +12 mm the front crosses the widest point and the frame can sit on the cheekbones rather than on the nose, a different plausible path to the same complaint. Both directions are consistent with sliding and only one is consistent with pinching, so the residual separates pinching from everything else, not sliding from pinching. That is a smaller claim than the one the keyword implies, and smaller than the one this page set out to test.
What this page cannot tell you
Four things are missing from the arithmetic above and none of them can be recovered from a published page. The distance from the inside edge of a front to its outside edge appears nowhere, which is why the internal ladder in the second table is left unconverted rather than converted with a guess. The corner geometry a hinge is mounted into is printed by none of the thirteen either, so the zero point of a hinge-to-hinge measure is a convention — this site reads the span as two lenses plus the bridge, the reading its own bridge page uses, and another manufacturer's definition, centre to centre at the hinge screws, is a subtly different line even under the same name. The hinge span in the first table is a computed quantity as well: it comes out of two stamped digits under that convention, so every column to its right inherits whatever the stamp leaves out, and the residuals in the last column are arithmetic, not caliper readings.
Two more gaps are the site's own, and this page is not going to describe the family's fault without describing it. The heart-shaped-face table and the oval-face table carry the same 130 to 145 mm under two different names, one calling it the front end piece to end piece and the other calling it the hinge span, which is the exact shape of the error the second table is built to catch; on the medians this site publishes, the two rows cannot both be about the same object, and neither has been corrected here, because a page that is already published does not get its numbers changed by an arithmetic page that points at it. And the allowance itself is a median from 3,942 records rather than a measurement of the frame in your hand, which is why the first table runs the conversion twice, at metal and at acetate, instead of naming one number as the truth. The earlier fitting guidance in the record puts the end piece at 3 mm per side, 6 mm across the frame, and that is a fifth of the acetate median; the sensitivity in the note is the honest response to a quantity the record disagrees about by a factor of three, not a hedge.
What is left is a procedure with one subtraction in it, and it is short. Measure your cheekbone line. Take the two stamped digits off the arm of any frame you are considering, add them as column two, double the lens and add the bridge as column three, and then decide which column the page's band is talking about by asking the one question that separates them: does its instruction put the caliper at the hinges or at the outside of the frame? If the answer is neither, the band is not on the same quantity as yours and the two numbers cannot be compared until it is moved. The material decides the rest, and the material is the only one of those inputs you can see from across a shop.
Thirteen published positions, six of them dated
The rows below are the positions this page's argument is built from, ordered by the date each carries rather than by whether they agree, and the seven that carry no date of their own sit at the bottom, since a position nobody dated cannot be sorted against positions somebody did. Of the six that carry a date, two are books whose own records give a year and nothing finer — 1850 and 1921 — which puts the span from the oldest position in the table to the retailer's measurements guide stamped 23 September 2026 at 176 years, longer than the stamped-size system has existed for most of it. Narrow the span to the positions that print a day as well as a year and four remain, from 2 April 2025 to 23 September 2026: 539 days, one year, five months and twenty-one days. Inside that, two pages carry the identical calendar day one year apart — 2 April 2025 and 2 April 2026 — and they are a good pair to read against each other, because the earlier one prints a width number with a tolerance and two incompatible measurement points, and the later one prints eight uses of the width word and not one millimetre of it.
The last column is the one that makes the table worth the reading, and across 176 years it barely moves. The 1850 treatise leaves you with three failure modes on three different widths and no dimensions for any of them. The 1921 book leaves you with adjectives about how wide a face reads, which is a different quantity from the one its title's subject has. The 2025 guide leaves you with a number and a two-millimetre window around it, and with a second sentence on the same page that moves the window 13 to 20 mm. The 2026 guide leaves you with a six-rung band and a face measurement spent on two other quantities. One position in thirteen, on a page whose subject is the width of a frame, admits that its own subject is not printed: “This measurement is not printed on the temple of your glasses.” Nobody in the whole table does the conversion between the rungs and the line on your head, which is the only claim this page is making.
Dates as each source carries them, read against the pages on 26 September 2026. Rows are ordered by date; the seven positions whose pages carry no date of their own sit at the bottom, and the count of six dated rows treats the two books as dated to their imprint years. Column three quotes each source verbatim except the two books, which are described in this page's words because their text was read through optical character recognition and is not clean. This site's own two width rows are in the second table rather than this one, marked there as its own, because a page that criticises a band for not naming its measurement point has to submit its own. No links are given to any of them.
| Published | What it prints | The width statement, as printed | What its own instruction leaves you holding |
|---|---|---|---|
| 1850, imprint year and nothing else | Three failure modes, on three different widths, none of them measured | A medical treatise on spectacles argues that the framework must be light and of proper dimensions: too large and broad, it loses its steadiness and the eyes are tired by the movement of the image; too narrow and heavy, the temples are compressed and there is aching near the eye; and in the same paragraph, the two glasses should be neither too near nor too far from each other, or double vision and other anomalies of vision may follow | One sentence with a mechanism and a consequence — attached to the distance between the lenses, the only one of those five widths that any frame stamps; the width the buyer is asked to choose carries only sensations |
| 1921, imprint year and nothing else | A theory of apparent width, with no line segment | A book on choosing spectacles argues that a lightly rimmed or rimless front makes the face read wider and a dark rim makes it read narrower, and puts arms carried over the ear and a low nosepiece in the same list of widening causes | Adjectives about a face, on a page about a frame — the word does duty for two quantities and the reader is left to notice |
| 2 April 2025 | A worked target, a tolerance, and two measurement points in one page | “This is the overall width from left to right outer edges of the frame itself.” … “shows how we measure our Frame Width, the caliper touches with two ends of the hinge behind.” And: “if the width is within two millimeters in either direction, it should still fit comfortably. So for a target frame width of 136, if you set the range to 134-138, the glasses will all fit you.” | A number, a window around it, and a second instruction that moves the window 13 to 20 mm — its own two sentences, nobody else's arithmetic |
| 15 July 2025 | A definition, a finger, and two face-measured gates | “The frame width is the total horizontal measurement of the front of the frames.” “The frame width should ideally be just slightly wider than your face, leaving about a finger's width of space between the arm of the temple and your head.” “If the widest part of your face measures under 130mm, choose a small pair of glasses.” “Be careful not to choose a frame width wider than your forehead.” | A definition that is on the object and a procedure that is on the face, joined by a finger; ten millimetre figures on the page, none of them on the sentence about pinching |
| 2 April 2026 | Eight uses of the word, twelve millimetres elsewhere, zero on the width | “We generally recommend that frame size should match face size and pupil position so the eyes land in an appropriate area of the lens.” The millimetres on the page belong to lens, bridge and arm, and to two lens-width suggestions sorted by narrow and average faces | A matching rule with two operands and no arithmetic — and the sizes it does give are lens sizes, which are not the width the sentence asks for |
| 23 September 2026, on a page also stamped 30 June 2023 | The only tabulated six-rung width ladder, and a face table that routes around it | Junior under 121 mm, Extra Small for adults under 121, Small 121–128, Medium 129–134, Large 135–145, Extra Large over 145. Beside it: face size measured temple to temple, mapped to lens width and to arm length — Narrow under 129 mm giving 50 mm and under and 135–140 mm, Medium 130–139 giving 51–55 and 140–145, Wide over 139 giving 56 and above and 150 or above | Forty-six millimetre figures, the most of any position read, and a face measurement spent on two other numbers while its own width ladder goes unconverted; no measurement point stated for either ladder |
| Undated at the width section | The admission, in the middle of a page that prints everything else | “You can usually find the measurements on the inside of the temple of your glasses.” Then, of the width: “This measurement is not printed on the temple of your glasses.” Its advice section for the width carries fourteen millimetre figures on other quantities and none here, only: “Find a frame that is slightly wider than your face — no more than a finger width between the temple and your face.” | The correct fact about the stamp, and a finger — a page that knows which of the five widths is unprinted and still cannot tell you its value |
| Undated at the definition | A measurement point that is a third one, named for the first time | “W – width is measured center to center at the temple hinge screws.” On the same page: “L – length refers to temple length, measured from the temple screw to the end of the temple tip,” and one table of per-model W, L and H values in millimetres | A line you could reproduce with a ruler, belonging to no other page in the table, and no statement of how far it sits from the hinge-to-hinge run the next row uses |
| Undated at the table | The only tabulated four-rung ladder, with its point printed in the header | “Frame Width Measure the frame horizontally from hinge to hinge to determine the total width.” Narrow 125–128 mm, Medium 129–132, Wide 133–136, Extra Wide 137–140 | A ladder you can stand on if you know your own hinge span — and nine millimetre figures, none of them on a head |
| Undated at the bands | Sizes by the lens, and a sentence about the front | Adult rungs of X-small under 46 mm, Small 47–49, Medium 50–53, Large 54–56, X-large 57 and above, every one of them a lens width: “Frame width: When the frame is roughly the same width as the face (not much wider, not much narrower), it tends to look more balanced and feel more secure.” Sport and wrap styles are the exception, “use the total frame width to determine size” | A size ladder in a different quantity from the word it shares a page with, and a hint that the total is used where the front curves — which is the one case the straight-line definitions cannot reach |
| Undated at the ladder | A fourth set of rungs, measured inside | Small under 125 mm, medium 126–130, large 131–135, extra-large 136 and above, on a page whose quantity is the internal frame width; nine millimetre figures | Rungs that cannot be compared to the other three ladders at all, because no page in the table prints the distance from an inside edge to an outside edge |
| Undated at the temple line | The word moved onto the face | “Frame width is the most straightforward aspect and is determined by measuring the distance between your temples.” No millimetre figure anywhere on the page | A definition that is a line on your head wearing the name of a line on the object, and no conversion to any band — which is the row that makes the card method in the next one legible |
| Undated at the size guide | A front defined as two numbers, and a face measured with a card | “Frame size is defined by lens and bridge width.” “MEASURE YOUR FACE WIDTH If you don't already own a pair of glasses with measurements, you can measure your face using a card.” Zero millimetre figures on the page, in either unit | An instruction that produces a face measurement the page's own definition of size has no column for — a number with nowhere to go |
Five widths, four names, one printed band. The digits on the inside of an arm give you two lenses and a bridge, from which a hinge-to-hinge span follows by addition and the outside of the front follows by adding the two end pieces — and the end pieces are the only term in the chain with no stamp, which is why three pages can each sell a rung called Medium without contradicting each other in words and contradict each other by 13 to 20 mm in the object. The conversion is one subtraction against one line on your own head, and the sensitivity of that subtraction is a material you can see and cannot order by number.
Two things would change what is written here. If any page in this family published the distance from the inside edge of a front to its outside edge, the third row of the second table would stop being an exception and would enter the same arithmetic as the rest; and if this site's own two width tables are reconciled — the 130 to 145 mm that appears once as a hinge span and once as a front, at two points the published medians separate by more than the tolerance both pages allow — then the last two rows of that table become one row and one of the two pages is wrong. That is a better outcome than the symmetry this page has had to leave in place.
The bands are not worthless and the pages are not lying. They are describing four different lengths with one word, on the assumption that a shopper who has never held a caliper cannot tell them apart and will not be told. Both assumptions hold. Take the cheekbone measurement first, write that number down, and when a size you like arrives in a rim of acetate rather than metal, look at it again: that is the term the arithmetic was missing, and it is the one that decides whether the same digits were ever the same fit.