Metric to Imperial Fastener Cross-Reference Calculator

The closest thread in the other system, with the diameter and the pitch differences both printed and the engagement depth at which the two forms walk apart — because nothing here is equivalent to anything, and the near misses are the dangerous ones.

Last updated 24 Sep 2026

Metric / imperial fastener cross-reference

A thread → its nearest neighbour, and how far off

Which way roundThe page always answers with the CLOSEST thread in the other system and then says how far away it is. It never says equivalent, because nothing here is: the two systems were built on different units from different reference profiles and no size in one is a size in the other.

Metric threadCoarse pitches from ISO 262’s selection plus the fine pitches that matter for this question. The pitch is part of the designation for a reason: M8 alone is ambiguous, and an M8 × 1 will start into an M8 × 1.25 nut and strip it.

Unified inch threadUNC and UNF from ASME B1.1. Gauge numbers below a quarter inch: the major diameter is 0.060 + 0.013·n inches, so a #10 is 0.190 in and 4.826 mm. UNF is not simply ‘fine’ — on some sizes the UNF pitch is closer to a metric coarse pitch than the UNC one is.

Metric property classThe designation is arithmetic: class a.b has a nominal tensile strength of 100a N/mm² and a nominal yield of 10ab. 8.8 is 800 and 640. ISO 898-1 raises the specified yield above that nominal for 8.8 over 16 mm, for 10.9 and for 12.9, so the rule is a floor rather than an identity. The stainless classes are ISO 3506 and are a different kind of thing — see the note below.

SAE grade to compare it withSAE J429, in ksi converted at 6.894757 N/mm² per ksi. Watch the size ranges: grade 5 drops from 85 ksi proof to 74 above one inch and grade 2 from 55 to 33 above three quarters, so ‘grade 5’ is two different materials depending on the bolt. Comparison charts almost always print only the small-size row.

Not a circuit: two thread forms interlocked, drawn with the pitch difference exaggerated to six per cent so that nine threads fit on a page. At the left-hand tick the two are in step: the crests sit in the roots and the flanks are in contact, the way a bolt sits in a nut. Each turn they slip by the pitch difference, and by the right-hand tick they have drifted half a pitch — crest meeting crest, nothing in contact at all. That is why a mismatched pair does not fail on the first turn: it starts easily, feels normal, and stops holding somewhere down the hole. The live scale underneath is the other half of the story and the one that usually decides it: the MAJOR DIAMETER gap between your thread and the closest one in the other system, on a scale of nought to 0.6 mm, with the same gap expressed in thread heights beside it. A gap of much under half a thread height is the dangerous region, because the smaller bolt will enter the larger nut and hold on almost nothing.

0.174mmExample

An M5 × 0.8, with class 8.8 against SAE grade 5

Two differences, and the thread count that turns them into a failure

Δd = |d₁ − d₂| · ΔP = |P₁ − P₂| · n_half = 0.5 × 0.5413·P / ΔP · P = 25.4 / TPI

Δd: the major diameter difference, in millimetres and as a percentage. It decides whether the pair will START. A bolt bigger than the nut’s thread will not enter; a bolt smaller than it enters freely and engages on less flank, which is the direction that assembles and fails
ΔP: the pitch difference. It decides whether the pair, having started, will keep engaging. It accumulates: after n threads the two forms are n·ΔP out of step
0.5413·P: the ISO 68-1 internal thread height, 5H/8 with H = P√3/2. Exactly 5/8 × √3/2 = 0.5412659, not the 0.54125 that is often printed. It is the depth the flanks have to share
n_half: how many threads of engagement it takes for the accumulated pitch error to reach HALF a thread height, at which point the flanks have lost each other and only the threads nearest the mouth carry anything. Under about three, the pair will not start. Over about ten, it starts easily and strips somewhere down the hole, which is worse
25.4 / TPI: threads per inch to pitch in millimetres, and the reason a #10-32 and an M5 × 0.8 look alike: 25.4/32 = 0.79375 against 0.8, which is 0.79 per cent. The pitch ↔ TPI arithmetic and the standard series belong to the converters section of this site; this page is about what happens when you try to assemble the pair
A_s: the tensile stress area, on each standard’s own coefficient: π/4(d − 0.938194P)² for metric and π/4(d − 0.974279P)² for unified. They are different because the two standards take the area at different points in the profile, which is a small but real reason the strength comparison is never exact

Worked example

An M5 × 0.8, with class 8.8 against SAE grade 5
THE NEAREST UNIFIED THREAD to an M5 × 0.8 is a #10-32 UNF, found by minimising the WORSE of the two percentage differences rather than either one alone — a pair that matches on diameter and misses on pitch is exactly as useless as the reverse
THE DIAMETERS. A #10 is 0.060 + 0.013 × 10 = 0.190 inches = 4.826 mm. An M5 is 5.000. The gap is 0.174 mm, or 3.61 per cent
THE PITCHES. 32 threads per inch is 25.4/32 = 0.79375 mm against the M5's 0.8. That is 0.00625 mm, or 0.79 per cent — which is inside the pitch tolerance of a commercial thread class, and is the reason this pair is famous
SO WHICH ONE KILLS IT? One thread height here is 0.5413 × 0.79375 = 0.4296 mm. The pitch error takes 34.4 threads to reach half of that, so on pitch alone the pair would run the full depth of any ordinary nut. The DIAMETER gap of 0.174 mm is 0.40 of a thread height straight away, at the first turn. This is a diameter mismatch, not a pitch one, and the brief this page was written from had it the other way round
AND THE ASYMMETRY IS THE WHOLE SAFETY POINT. An M5 bolt is BIGGER than a #10-32, so it will not enter a #10-32 nut and you find out immediately. A #10-32 bolt is SMALLER than an M5, so it drops into an M5 nut, spins in by hand, feels like a thread and engages on a sliver of flank. The combination that goes together is the combination that fails, and it fails at a load nobody predicted because the stress area is right and the engagement is not
THE STRENGTHS, for completeness. An M5's tensile stress area is 14.18 mm² and a #10-32's is 12.90 mm². At class 8.8 the metric bolt's proof load is 580 × 14.18 = 8.23 kN; at SAE grade 5 the inch bolt's is 586.1 × 12.90 = 7.56 kN. The materials are within about one per cent of each other on proof stress and the BOLTS are 8.8 per cent apart, because the areas differ. Equivalent materials do not make equivalent fasteners
THE VERDICT, and it is the same verdict for every row in the table below. These are two different threads. Mixing them in a load-bearing joint is a defect and not a workaround: it is not inspectable after assembly, it passes a torque check, and it fails at a fraction of the design load with no warning

Every metric/unified pair within six per cent on both diameter and pitch

Computed from the two series rather than quoted, which is why the order may not be the one you expect. The tightest pair of all is M2.5 against a #3-56 UNF: 0.58 per cent on diameter and 0.79 on pitch, which is inside the tolerance class on both, and the reason it is not famous is that almost nobody uses a #3-56. The famous one, M5 × 0.8 against #10-32 UNF, is a DIAMETER mismatch and not a pitch one — the pitches are within 0.79 per cent and would run for thirty-four threads, but the major diameters are 0.174 mm apart, which is 0.40 of a thread height. So the failure is asymmetric and that is the useful part: an M5 bolt will not enter a #10-32 nut at all, being too big, while a #10-32 bolt drops into an M5 nut, spins in easily and holds on a sliver of flank. The one that goes together is the one that fails. Read the last two columns together: a long thread-count with a small diameter gap is a joint that will assemble and will not hold. These dimensions come from a published standard’s table, not from a formula. The standard itself is cited below and the printed values are attributed to the catalogue they were taken from; a different publisher may round differently in the last digit. This page sizes a part; it does not certify one. Where the answer carries a consequence — a load path, a lifting duty, a pressure boundary, a fastener holding something that can fall — confirm it against the design code that governs the application, and against the manufacturer’s own rating, before relying on it.
Metric	Unified	Diameter gap (mm)	…%	Pitch gap (mm)	…%	Threads to half a thread height	What happens
M2.5	#3-56 UNF	-0.015	0.58	-0.0036	0.79	34.1	WILL start and WILL strip. The dangerous one
M5	#10-32 UNF	0.174	3.61	0.0063	0.79	34.4	pitches match, diameters do not — will not start
M4	#8-36 UNF	-0.166	3.98	-0.0056	0.79	34.1	pitches match, diameters do not — will not start
M20	3/4-10 UNC	0.950	4.99	-0.0400	1.57	16.9	binds within a few threads
M18	3/4-10 UNC	-1.050	5.51	-0.0400	1.57	16.9	binds within a few threads
M10	3/8-16 UNC	0.475	4.99	-0.0875	5.51	4.6	binds within a few threads
M24	1-8 UNC	-1.400	5.51	-0.1750	5.51	4.6	binds within a few threads
M12x1.25	1/2-20 UNF	-0.700	5.51	-0.0200	1.57	16.9	binds within a few threads
M20x1.5	3/4-16 UNF	0.950	4.99	-0.0875	5.51	4.6	binds within a few threads
M3	#4-48 UNF	0.155	5.46	-0.0292	5.51	4.6	binds within a few threads
M14	9/16-12 UNC	-0.287	2.01	-0.1167	5.51	4.6	binds within a few threads
M8x1	5/16-24 UNF	0.063	0.79	-0.0583	5.51	4.6	diameters within one and a half per cent; it will start
M10x1	3/8-24 UNF	0.475	4.99	-0.0583	5.51	4.6	binds within a few threads
M3.5	#6-40 UNF	-0.005	0.15	-0.0350	5.51	4.6	diameters within one and a half per cent; it will start

Metric property class against SAE grade, with the gaps printed

The two commonest equivalences on earth, 8.8 ≈ grade 5 and 10.9 ≈ grade 8, and neither is an equality. Look at the first row’s three gap columns and notice that they do not have the same sign: SAE grade 5 is one per cent ABOVE class 8.8 on proof stress and three and a half per cent above on ultimate, while class 8.8 is nine tenths of a per cent above grade 5 on yield. Neither is simply stronger; they cross over between the three measures. For 10.9 against grade 8 the proof stresses are within a third of a per cent and the yields nearly five per cent apart. That is close enough for the pairing to be a useful shorthand and not close enough for a substitution on a drawing, because a substitution changes the thread as well as the material and the thread is the part that will not go together. The last row is the one people get wrong most expensively. A2-70 stainless is weaker than either — its 0.2 per cent proof strength is 450 N/mm² against class 8.8’s 640, thirty per cent down — and it has NO proof load stress at all, because ISO 3506 does not define one. There is nothing to put in that column, which is a more fundamental problem than a low number: the quantity every torque table is built on does not exist for it. Take the torque figures from the stainless maker, not from a steel table scaled down, and remember that austenitic stainless galls. These dimensions come from a published standard’s table, not from a formula. The standard itself is cited below and the printed values are attributed to the catalogue they were taken from; a different publisher may round differently in the last digit. The same designation can mean different dimensions in different standards families — ANSI against ISO, inch against metric, one national standard against another. The family used here is named beside every figure; check which one your part was made to. This page sizes a part; it does not certify one. Where the answer carries a consequence — a load path, a lifting duty, a pressure boundary, a fastener holding something that can fall — confirm it against the design code that governs the application, and against the manufacturer’s own rating, before relying on it.
Metric	SAE	Metric proof (N/mm²)	SAE proof (N/mm²)	Gap %	Metric yield	SAE yield	Gap %	Metric ultimate	SAE ultimate	Gap %
ISO 898-1 class 8.8 (d ≤ 16)	SAE grade 5, ≤ 1 in	580	586.1	-1.03	640	634.3	0.90	800	827.4	-3.31
ISO 898-1 class 10.9	SAE grade 8	830	827.4	0.32	940	896.3	4.87	1,040	1,034.2	0.56
ISO 898-1 class 9.8 (d ≤ 16)	SAE grade 8	650	827.4	-21.44	720	896.3	-19.67	900	1,034.2	-12.98
ISO 898-1 class 5.8	SAE grade 2, ≤ 3/4 in	380	379.2	0.21	400	393.0	1.78	520	510.2	1.92
ISO 898-1 class 4.6	SAE grade 1	225	227.5	-1.11	240	248.2	-3.31	400	413.7	-3.31
ISO 3506-1 A2-70 / A4-70 (class 70)	— nothing comparable	none defined	—	—	450	—	-29.7 against 8.8	700	—	-12.5 against 8.8

Across-flats: every metric spanner against its nearest imperial one

Sorted by how close the pair is, which puts the two that the usual charts name a long way down the list. The tightest of all is 27 mm against 1-1/16 inch at 0.0125 mm — twelve microns, four times closer than the famous 19 against 3/4 inch — and the second is 46 mm against 1-13/16 at 0.0375. Those are interchangeable in practice and nothing bad happens. The 19 mm against 3/4 inch pair the charts do name is real, at 0.05 mm, but notice WHICH metric bolt has a 19 mm head: it is the DIN 931/933 M12, not the ISO 4014 one, which is 18 mm. ISO 272:1982 changed the across-flats on exactly four sizes — M10, M12, M14 and M22 — and DIN did not follow, so whether you have a near-miss depends on which standard your bolt was made to. The pairs that do damage are the 0.1 to 0.3 mm ones: 16 mm against 5/8 inch at 0.125, 24 against 15/16 at 0.1875, 13 against 1/2 inch at 0.30. At that gap the spanner goes on, feels wrong to nobody, bears on the corners of the hexagon instead of the flats, and rounds the head at the torque where it matters. Note the sign convention: a NEGATIVE gap means the imperial spanner is smaller and simply will not go on — which is the safe direction — while the metric spanner is then the loose one on the imperial head. These dimensions come from a published standard’s table, not from a formula. The standard itself is cited below and the printed values are attributed to the catalogue they were taken from; a different publisher may round differently in the last digit. This page sizes a part; it does not certify one. Where the answer carries a consequence — a load path, a lifting duty, a pressure boundary, a fastener holding something that can fall — confirm it against the design code that governs the application, and against the manufacturer’s own rating, before relying on it.
Metric (mm)	Fits	Imperial	…in mm	Fits	Imperial minus metric (mm)	…%	What happens
27.0	M18 (ISO 4014)	1-1/16 in	26.9875	5/8 in heavy	-0.0125	-0.05	fits both ways; you will never know which you picked up
46.0	M30 (ISO 4014)	1-13/16 in	46.0375	1-1/8 in heavy	0.0375	0.08	fits both ways; you will never know which you picked up
19.0	M12 (DIN 931/933)	3/4 in	19.0500	1/2 in	0.0500	0.26	fits both ways; you will never know which you picked up
16.0	M10 (ISO 4014)	5/8 in	15.8750	7/16 in	-0.1250	-0.78	imperial will not go on the metric head; the metric spanner is loose on the imperial one
24.0	M16 (ISO 4014)	15/16 in	23.8125	5/8 in	-0.1875	-0.78	imperial will not go on the metric head; the metric spanner is loose on the imperial one
22.0	M14 (DIN 931/933)	7/8 in	22.2250	1/2 in heavy	0.2250	1.02	imperial is LOOSE on the metric head — rounds it
32.0	M22 (DIN 931/933)	1-1/4 in	31.7500	3/4 in heavy	-0.2500	-0.78	imperial will not go on the metric head; the metric spanner is loose on the imperial one
41.0	M27 (ISO 4014)	1-5/8 in	41.2750	1 in heavy	0.2750	0.67	imperial is LOOSE on the metric head — rounds it
13.0	M8 (ISO 4014)	1/2 in	12.7000	5/16 in	-0.3000	-2.31	imperial will not go on the metric head; the metric spanner is loose on the imperial one
17.0	M10 (DIN 931/933)	11/16 in	17.4625	3/8 in heavy	0.4625	2.72	imperial is visibly loose — do not
36.0	M24 (ISO 4014)	1-7/16 in	36.5125	7/8 in heavy	0.5125	1.42	imperial is visibly loose — do not
18.0	M12 (ISO 4014)	11/16 in	17.4625	3/8 in heavy	-0.5375	-2.99	imperial will not fit at all
34.0	M22 (ISO 4014)	1-5/16 in	33.3375	7/8 in	-0.6625	-1.95	imperial will not fit at all
50.0	M33 (ISO 4014)	2 in	50.8000	1-1/4 in heavy	0.8000	1.60	imperial is visibly loose — do not

Nothing here is equivalent, and the near misses are the ones that get built

Nothing on this page is equivalent to anything, and that is the point. The metric and unified series were built on different units, from different reference profiles, by different committees, and no size in one is a size in the other. What this page does instead is find the CLOSEST thread in the other system and print how far away it is — in millimetres and in per cent, on the major diameter and on the pitch separately, because the two failures are different. The diameter decides whether the pair will start. The pitch decides whether, having started, it will keep engaging. A pair can be excellent on one and hopeless on the other, and the M5 against #10-32 case is exactly that.

Why the near misses are the whole subject. A pair that is twenty per cent apart is harmless: it will not go together, you find out in a second, and nothing is damaged. A pair that is one per cent apart is a defect waiting for a load. The smaller bolt enters the larger nut, turns by hand, takes a torque wrench reading that looks right — because most of the torque is friction under the head and in the thread, not flank contact — and holds on a sliver of engagement. Nothing about it is visible once it is assembled. When it goes, it strips rather than breaks, so the joint opens instead of cracking in place. The table on this page lists every metric/unified pair within six per cent on both measures, computed from the two series rather than quoted, and there are not many of them: the tightest is M2.5 against a #3-56 UNF at 0.58 and 0.79 per cent.

What the pitch arithmetic is, and where it lives. Pitch in millimetres is 25.4 divided by threads per inch, and the standard series and that conversion are the subject of a page in the converters section of this site rather than of this one. This page starts where that one ends: given that an M5 × 0.8 and a #10-32 UNF have pitches 0.79 per cent apart, what actually happens when you try to assemble them? The answer is the thread count. Each turn, the two forms slip by the pitch difference; after n threads they are n·ΔP out of step; and once that reaches half a thread height (5H/8 = 0.5413·P) the flanks have lost each other and only the threads nearest the mouth are carrying anything. Under about three threads the pair will not start at all. Over about ten it starts easily and strips deep in the hole, which is worse, because it gets assembled.

The strength equivalences, and where they stop being true. Class 8.8 against SAE grade 5 and class 10.9 against grade 8 are genuinely close — 10.9 and grade 8 are within a third of a per cent on proof stress — but the gaps do not have the same sign on every measure. Grade 5 is about one per cent above 8.8 on proof stress and three and a half per cent above on ultimate, while 8.8 is a shade above grade 5 on yield. Neither is simply stronger. And the comparison is between MATERIALS: the bolts also differ in tensile stress area, both because the diameters differ and because the two standards define the area with different coefficients (0.938194 for metric, 0.974279 for unified), so equivalent materials do not give equivalent proof loads. A2-70 stainless is the one that surprises people: it is thirty per cent below class 8.8 on yield, and ISO 3506 defines no proof load stress for it at all, so the quantity every torque table is built on does not exist for the material.

Mixing the two systems in a load-bearing joint is a defect. Not a compromise, not a field expedient, not something to note and move on from. The reasons are cumulative. The engagement is a fraction of what the design assumed and cannot be measured after assembly. The joint passes every check that can be applied to it. The failure mode is stripping, which is sudden and complete rather than progressive. And on a tapped hole it is not recoverable: a stripped female thread has to be drilled out and inserted or the part scrapped. The same applies to a spanner: an imperial spanner two tenths of a millimetre loose on a metric head bears on the corners instead of the flats and rounds them at the torque where the joint mattered, and a rounded head has no second chance either. Pipe threads deserve the same sentence for a different reason — a G ½ and a ½ NPT share fourteen threads per inch and differ by 1.84 per cent on diameter and by five degrees of flank angle, 55 against 60, so the male will enter and will not seal.

Frequently asked questions

Is an M5 the same as a #10-32?

No. The major diameters are 5.000 and 4.826 mm, which is 0.174 mm and 3.6 per cent apart; the pitches are 0.800 and 0.79375 mm, which is 0.79 per cent apart. So the pitches are close enough to be inside a commercial thread class and the diameters are not, and the pair fails on diameter at the first turn. The asymmetry matters: an M5 bolt is too big to enter a #10-32 nut and you find out at once, while a #10-32 bolt drops into an M5 nut, turns freely and holds on almost nothing. The combination that assembles is the one that fails.

Is class 8.8 the same as SAE grade 5?

Close, and not the same. On proof stress grade 5 is about one per cent higher (586 against 580 N/mm²); on ultimate it is three and a half per cent higher; on yield class 8.8 is very slightly higher. The gaps do not all run the same way, which is why the pairing is a shorthand rather than an equivalence. Two further cautions. SAE grade 5 falls from 85 to 74 ksi proof above one inch, and comparison charts nearly always print only the small-size row. And the comparison is between materials: the bolts differ in stress area as well, so equal materials do not give equal proof loads.

How does A2-70 stainless compare with 8.8?

It is weaker on both measures that can be compared, and on the measure that matters most there is nothing to compare. A2-70’s 0.2 per cent proof strength is 450 N/mm² against class 8.8’s specified yield of 640, about thirty per cent down, and its minimum tensile strength is 700 against 800. ISO 3506 defines no proof load stress at all, so the number every torque table is built on does not exist for it. Take torque figures from the stainless maker rather than scaling a steel table, and allow for galling: austenitic stainless seizes on itself, the friction rises during tightening and a seized fastener is destroyed on removal.

Will a 13 mm spanner fit a 1/2 inch bolt?

Yes, and it will damage it. Half an inch is 12.70 mm, so a 13 mm spanner has 0.30 mm of slack — enough that it bears on the corners of the hexagon rather than on the flats, and the corners yield at the torque where you needed the joint. It is worth being clear about the direction, because the usual charts state it backwards: the METRIC spanner is the loose one here. A half inch spanner will not go onto a 13 mm head at all, which is the safe failure. The genuinely interchangeable pairs are much closer than either: 27 mm against 1-1/16 inch is 0.0125 mm apart and 46 against 1-13/16 is 0.0375.

What about 19 mm and 3/4 inch?

Three quarters of an inch is 19.05 mm, so the pair is 0.05 mm apart and is effectively interchangeable. The detail the charts leave out is which metric bolt has a 19 mm head: it is the DIN 931/933 M12, not the ISO 4014 M12, which is 18 mm. ISO 272:1982 changed the across-flats on four sizes — M10, M12, M14 and M22 — and the DIN products kept the older figure, so whether you have a near-miss depends on which standard your bolt was made to and not just on its thread.

How many threads of engagement does a mismatched pair get?

Divide half a thread height by the pitch difference. One thread height is 0.5413 × pitch, so for two threads 0.79 per cent apart at a 0.8 mm pitch that is about thirty-four threads — deeper than any ordinary nut, which is why such a pair assembles happily if the diameters allow it. For two threads 5.5 per cent apart it is under five, so the pair binds within a few turns and the mistake is caught. Counter-intuitively the closer pair is the more dangerous one, because it is the one that gets built into something.

Can I use a metric bolt in an imperial tapped hole in an emergency?

In a load-bearing joint, no. The engagement is a fraction of what was designed, nothing about it is visible after assembly, it will pass a torque check because most of the torque is friction rather than flank contact, and the failure mode is stripping rather than breaking — so the joint opens rather than cracking where an inspector could find it. On a tapped hole it is also not recoverable: the female thread is destroyed and the part needs drilling out and inserting or scrapping. If the hole is already damaged, a thread insert in the correct size is the repair; a bolt from the wrong system is not.

Related calculators

References

ISO 261:1998, ISO general purpose metric screw threads — General plan, and ISO 262 for the selected sizes; ASME B1.1 for the unified inch series. Both cited by number and neither reproduced. The diameter and pitch pairs used here are the ones already verified in the CONVERTERS batch that built thread-pitch-and-tpi-converter, from Engineering ToolBox’s UNC and UNF charts, and they reproduce ISO 898-1’s published tensile stress area column to the last printed digit — which is a check on both numbers at once, since the area depends on d and P together.
ISO 898-1:2013, Mechanical properties of fasteners made of carbon steel and alloy steel — Part 1: Bolts, screws and studs with specified property classes. Copyrighted and cited by number. The property class figures used here are the ones this plugin’s bolt-proof-load page already carries, taken from Kova’s published extract of Table 3 and cross-checked against RoyMech’s table. The class designation is itself computable and that is the point: the first number is the nominal tensile strength in hundreds of N/mm², and the two multiplied give the yield.
SAE J429, Mechanical and Material Requirements for Externally Threaded Fasteners. Cited by number. The grade figures are the ones already verified in this plugin’s bolt-proof-load page, from STS Industrial’s table with Portland Bolt agreeing on every row it prints. The size dependence is the part that gets dropped in comparison charts: grade 5 falls from 85 to 74 ksi proof above one inch, and grade 2 from 55 to 33 above three quarters.
ISO 3506-1:2020, Fasteners — Mechanical properties of corrosion-resistant stainless steel fasteners — Part 1: Bolts, screws and studs with specified grades and property classes. Cited by number. The fact this page needs from it is a NEGATIVE one, and it is the reason A2-70 cannot be slotted into a strength chart beside 8.8 and grade 5: ISO 3506 specifies no proof load stress at all. There is nothing to compare, because the quantity the comparison is made on does not exist for that material.
ISO 272:1982, Fasteners — Hexagon products — Widths across flats, with ISO 4014 and ASME B18.2.1 for the products themselves. Cited by number. The across-flats figures are batch M02a’s, which verified four sizes where ISO 272:1982 changed the width and DIN 931/933 did not — M10, M12, M14 and M22 — against Wikipedia’s ISO 272 table, EKINSUN’s wrench chart and Würth Industry’s DIN/EN/ISO comparison. Those four are exactly the sizes where a metric–imperial spanner near-miss depends on WHICH metric bolt you have.
Hydraulic Insight, BSP Thread Size Chart: BSPP & BSPT Dimensions, and Valves Online, Explanation of common threads including BSP & NPT — the pair already used and cross-checked for the converters plugin’s thread page — with Nuoan’s G 1/2 vs 1/2 NPT comparison (read 29 September 2026) as a third reading of the one case named here. All three agree that a G ½ and a ½ NPT share 14 threads per inch and differ in flank angle, 55° against 60°, and that a G ½ male will enter a ½ NPT port for a limited distance with the flank contact and the sealing geometry both wrong.