The Working End

Ropework ยท a one-page reference

Twelve knots, and the way each one lets go

Every knot is a trade. It gives back less strength than the rope had, it either jams after loading or it doesn’t, and it holds in one fibre and creeps out of another. A reference that only shows you the finished shape has told you the easy half.

This is deliberately one page: the drawings, the numbers and the caveats in a single document, so it prints, and so it still works on a phone with no signal. Nothing here is sold, sponsored or affiliated. The about section says who wrote it and how.

Three things before the knots

  1. The knot is the weak point. A bend in a loaded rope is where it breaks. Plan on losing a quarter to a half of the rope’s rated strength, and never the other way round.
  2. Dressing and setting are part of the knot. The same tuck sequence, left crossed instead of parallel, is a different knot with different numbers. Pull it into shape before you trust it.
  3. The right knot in the wrong rope is the wrong knot. A bowline that has held for two centuries in hemp and nylon will roll out of slippery high-modulus line. Choose the knot after you know the fibre.

One

The five shapes every knot is made of

Knot instructions are unreadable until these five words are unambiguous. Everything below is built out of them and nothing else.

Rope terminology: standing part and working end, bight, crossing turn, round turn around a post, and a half hitch Standing part & working end under load you tie with this Bight doubled back Crossing turn crossed once Round turn one full wrap Half hitch a crossing turn round something
Reading the diagrams. A strand drawn across the front of another passes in front of it; a strand that disappears behind a post goes round the back. Where a rope changes colour it is a different rope. The standing part is the length that carries the load, the working end is the short end you tie with, and a bight is any doubled-back section of rope that has not yet crossed itself. Cross it once and it becomes a crossing turn, which is the raw material of every hitch on this page. A turn is one pass round an object, a round turn is a full wrap that can hold most of a load by friction alone before any knot is tied, which is why it is the first move in the safest way to tie a loaded rope to a ring.

Two

The twelve

Four families: a stopper stops a rope going somewhere, a loop makes a fixed eye, a hitch attaches a rope to an object, a bend joins two ropes. Almost every rope problem is one of those four, and knowing which one you have is most of the work.

01

Figure-eight knot

stopper

also: Flemish knot

Keeps a rope end from running out through a block, a fairlead or a grommet. An overhand knot does the same job in less rope, and then jams so hard it usually has to be cut out; the figure eight is the one that comes back.

  1. Make a crossing turn a hand's width from the end.
  2. Take the working end behind the standing part.
  3. Bring it down through the loop.
  4. Pull the standing part and the end apart until both lobes lie flat.

Worth knowing. The tail is the part people get wrong: leave ten rope diameters of it. A stopper with a short tail is a stopper that has already half untied.

How it fails. It rarely breaks and it rarely slips. What it does is jam. After a shock load in wet rope, work it loose by bending the lobes with your thumbs, not by hauling on the ends. Left loose and unset in stiff or slick line it can shake out, so dress it and pull it up hard.

02

Figure-eight on a bight

fixed loop

also: figure-eight loop

The loop to tie when the knot will be loaded hard and checked by somebody else. Its shape is legible from three metres away, right or wrong, which is why climbing and rescue work standardised on it rather than on stronger, subtler knots.

  1. Double the rope back on itself to make a bight.
  2. Tie a plain figure eight using the doubled rope.
  3. Dress it: every pair of strands lies parallel, nothing crosses.
  4. Set it by pulling all four parts, and leave a long tail.

Worth knowing. A crossed, untidy eight is measurably weaker than a dressed one, and it looks wrong at a glance. That legibility is a safety property, not an aesthetic one.

How it fails. It jams. That is the trade for its strength: after a serious loading you may be untying it with a marlinspike, or not at all. It also needs an end or the middle of the rope. If the line is already round a tree you need a rethreaded eight or a bowline instead.

03

Bowline

fixed loop

also: king of knots, and oversold as such

A loop that will still untie after it has been loaded, which is why it survived four centuries of sail. Round a post, through a ring, into a sail's clew, onto a bucket handle.

  1. Make a small crossing turn in the standing part, loop uppermost.
  2. Bring the working end up through the loop.
  3. Round behind the standing part.
  4. Back down through the loop, alongside the part that came up.
  5. Set it by pulling the standing part against the loop.

Worth knowing. Fix it, don't avoid it: a long tail, a stopper in the tail, or a Yosemite finish. In high-modulus line, don't use it at all.

How it fails. Three ways, all documented. Pulled sideways across the loop (ring loading) it can capsize into a slip knot. Cycled loose and loaded, loose and loaded, the tail creeps out. In stiff or slippery rope it rolls out entirely. None of these are hypothetical; all three have killed people who trusted a bowline in the wrong rope.

04

Alpine butterfly

mid-line loop

also: lineman's loop

A loop in the middle of a rope, tied without either end, that takes a pull in all three directions at once. It is the knot for isolating a damaged section, for clipping the middle of a line, and for the loop in a trucker's hitch.

  1. Take two turns round your flat hand.
  2. Lift the turn nearest your fingertips over the other two and back toward your wrist.
  3. Pull that turn out from under the remaining turns to form the loop.
  4. Slide it off and set it by pulling both standing parts.

Worth knowing. Unlike an overhand loop, it does not jam after being loaded from two ends, which is why it is worth the extra ten seconds.

How it fails. Tied wrong it dresses into a shape that looks similar and pulls out of line under a three-way load. Check the finished knot: it is symmetrical, and the two standing parts leave the knot on opposite sides. It also deforms slightly before it grips when the pull is loop-only in soft rope.

05

Clove hitch

hitch

also: builder's hitch, ratline hitch

Fast, adjustable, and tieable in a bight with no end free. Fenders on a rail, a line onto a post while you work out what you actually want.

  1. Take a turn round the post.
  2. Take a second turn, crossing over the first diagonally.
  3. Tuck the working end under that second turn.
  4. Set both parts and check the two turns lie snug against each other.

Worth knowing. Back it up with a half hitch on the standing part and it becomes a reasonable hitch. On its own it is a way to hold something still, not a way to hold something safe.

How it fails. It slips. Under a load that changes direction, under vibration, or on a smooth or tapered post it walks and then lets go, and it can bind so hard the other way that it is hard to release under tension. Treat it as a temporary hitch or a component of a bigger one, never as the only thing holding a load.

06

Round turn and two half hitches

hitch

also: the ring hitch that actually works

The everyday way to make a loaded rope fast to a ring, rail or bollard. The round turn carries most of the load by friction before the knot exists, so you can tie the hitches with slack in your hands, and ease the load out again under control.

  1. Take a full round turn round the object, so the rope crosses its own path.
  2. Make a half hitch round the standing part.
  3. Make a second half hitch the same way round.
  4. Set the hitches down against the round turn.

Worth knowing. On a permanent mooring, seize the tail to the standing part with light line. Months of small movements will otherwise work it loose.

How it fails. The failure is a tying error: the second half hitch taken the opposite way round produces a cow hitch shape that creeps along the standing part under load. Both hitches must run in the same direction, so the finished pair looks like a clove hitch on the standing part.

07

Midshipman's hitch

adjustable hitch

the taut-line family, and the one to prefer

Tension a line by hand and have it stay: push the hitch toward the anchor to slacken, pull it back to tighten, and the load holds it in place. Guy lines, tarp ridge lines, a line over a load.

  1. Pass the end round the anchor and bring it back alongside the standing part.
  2. Take two turns round the standing part, both inside the loop, working toward the anchor.
  3. Take one more turn outside the loop.
  4. Dress the turns snug and load it to set.

Worth knowing. The variant with both turns inside the loop is the midshipman's; the common taut-line has them arranged so the hitch sits against the load and is more prone to rolling. Same family, better behaviour.

How it fails. It grips by friction, so it slips in wet, icy, dirty or high-modulus line, and it slips if the turns are loose. Under a sudden shock it either jams or runs, depending on where the turns sit. It is a hand-tension knot, not a safety knot.

haul, then lock offfixed endanchor
08

Trucker's hitch

tensioning system

also: dolly, waggoner's hitch

The way to pull a load down hard with rope you already have: a loop in the standing part acts as a turning point, and the end reeved through it gives you leverage no hand can match on the rope alone.

  1. Make the line fast at one end.
  2. Tie a loop in the standing part above the anchor: an alpine butterfly, not a slipped overhand.
  3. Pass the working end round the anchor and back up through the loop.
  4. Haul on the end. The theoretical gain is 3:1.
  5. Lock it off with two half hitches round both parts below the loop.

Worth knowing. Use a butterfly for the loop. An overhand-based loop jams under exactly the tension the hitch is designed to create, and you will be cutting it off.

How it fails. The hazard is the force, not the knot. Rope running over rope is a bad pulley: friction eats most of the gain, and hauled systems commonly measure nearer 2:1. But 2:1 on a body weight is still enough to crush a roof box, bow a rack rail, or take a thin cord past its rating. Everything in the system, including the anchor, must be rated for what your arms can generate.

thick ropethin rope
09

Sheet bend

bend

also: weaver's knot, becket bend

Joins two ropes, and unlike most bends it does not mind that they are of different diameter or stiffness. The thick rope forms the bight; the thin rope does the work.

  1. Make a bight in the thicker rope.
  2. Bring the thin rope up through the bight.
  3. Pass it round behind both legs of the bight.
  4. Tuck it under its own standing part, not under the bight.
  5. Check both short ends finish on the same side.

Worth knowing. Ends on opposite sides is the left-handed sheet bend: it looks nearly identical, holds worse, and is the single most common way this knot is tied wrong.

How it fails. It spills. With no load on it, a sheet bend shaken about will work itself apart, and it is among the weaker bends when it does hold. With a large difference in diameter, take a second round turn (the double sheet bend) or it rolls off the thick rope's bight.

10

Double fisherman's

bend

also: grapevine bend

The compact, symmetrical way to join two ropes of similar diameter, and the standard way to close a length of cord into a loop for friction hitches. Its low profile is why it passes over edges and through gear without snagging.

  1. Lay the ropes side by side, ends opposed.
  2. With one end, take two wraps round both ropes and thread it back through the wraps.
  3. Repeat with the other end, the mirror image of the first.
  4. Pull the two barrel knots together until they seat as one.

Worth knowing. Both barrels must be mirror images. If one is tied the other way round the pair sits crooked and seats badly, and checking for it takes a second.

How it fails. It does not come undone; that is the failure. After a heavy load, budget on cutting it rather than untying it, and never use it on a rope you will need to separate. In slippery high-modulus cord, use three wraps instead of two.

host ropecord
11

Prusik hitch

friction hitch

the classic of a large family

A cord that grips a rope when it is loaded and slides freely when it is not. Ascending a fixed line, backing up a lowering, hauling systems, holding a tensioned line while you re-tie the end.

  1. Girth-hitch the cord loop round the host rope.
  2. Pass the loop through itself twice more, so three wraps lie neatly side by side.
  3. Dress the wraps flat, because crossed wraps hold badly.
  4. Test it under body weight before it matters.

Worth knowing. Diameter is the whole trick: cord at roughly 60-80% of the host rope's diameter. Too close and it slips, too thin and it damages the sheath before it stops you. Four wraps in wet or icy conditions.

How it fails. It slips, then it melts. A prusik that runs under load glazes the cord, and the melting point of the cord is the real limit: roughly 215-260 °C for nylon, and about 145 °C for high-modulus polyethylene, which is why HMPE cord must never be used for a friction hitch.

12

Constrictor knot

binding

also: the knot you tie to cut off

Closes a sack, clamps a bundle, holds a hose while you work on it, and stops a rope end unlaying before you whip it. It tightens on itself and does not let go.

  1. Take a turn round the bundle.
  2. Cross a second turn over the first, as for a clove hitch.
  3. Tuck the end under the crossing rather than under the last turn alone.
  4. Pull both ends hard; it will not need checking again.

Worth knowing. For a rope end about to be whipped or a bundle that is going nowhere, nothing else is as quick or as certain.

How it fails. The trouble with it is the opposite of failure. On a thin-walled tube it can crush what it is holding, and once set in small cord it is effectively permanent. Tie it where you are content to cut it off, and use a slipped version if you want any chance of undoing it.

The reef knot is not a bend

A reef knot (two ropes, left over right, then right over left) is a binding knot. Tied round a bundle, with both parts pulling the same way, it is fine and has been for four thousand years: reefing a sail, tying a bandage, closing a parcel. Used to join two ropes that pull against each other, it capsizes into a pair of half hitches and runs, especially when the two ropes differ in diameter or when one end is pulled sideways. It has drowned and dropped enough people to have its own literature. Join ropes with a sheet bend or a double fisherman's; keep the reef knot for parcels.

The same knot tied with an extra half turn, left over right and then left over right again, is the granny, which is worse in every respect: it slips sooner, jams harder, and looks almost identical. If your reef knot does not lie flat and square, that is what you have tied.

Three

What a knot costs you

A loaded rope does not break in the straight run. It breaks where it bends tightest, because the fibres on the outside of that bend carry a larger share of the load than the ones on the inside, and they reach their limit first. Anything that eases that bend gives strength back: a larger radius, more turns to spread the load, a splice in place of a knot.

0%25%50%75%100%Eye splice, well made90-100%Figure-eight loop70-80%Double fisherman's65-80%Alpine butterfly65-75%Bowline65-75%Round turn & two half hitches60-75%Clove hitch60-70%Overhand loop55-65%Sheet bend45-60%Reef knot used as a bend40-50%
Published efficiency ranges, as a percentage of the rope’s rated breaking strength. These are collected from break-test literature rather than measured here: the sailing-terminations work of Milne and McLaren, the rescue and climbing community’s pull-test series (Moyer’s tests and RopeLab’s published results among them), and the Cordage Institute’s splice figures. They disagree with each other, and they should, because efficiency moves with rope construction, diameter, dressing, wet or dry, and how fast the load came on. The ordering is stable across studies even where the values are not, and the ordering is the part worth memorising.

How to use these numbers

Not as design values. A rope’s working load limit comes from its manufacturer, and it already includes a safety factor, commonly around 5:1 for general rigging and 10:1 or more where a person is on the end. Knot efficiency is a further reduction on top of that, and the sensible way to use it is as a reason to prefer one knot over another, not as a licence to calculate how close to the edge you can work.

Dynamic loads are the part that catches people out. A weight dropped even a short distance onto a static line generates forces several times its own weight, which is why a low-stretch rope in a fall arrest role is dangerous in a way its breaking strength does not reveal. Stretch is not a weakness; it is how the rope absorbs energy that would otherwise go into the anchor, the knot, or you.

The four things that move the number

  • Bend radius. Tight bends concentrate strain. This is why a splice, which has no tight bend at all, keeps almost everything, and why a knot round a thin shackle pin is worse than the same knot round a fat bollard.
  • Dressing. Crossed strands inside a knot bear unevenly. A tidy knot and an untidy one of the same name are different knots by several per cent.
  • Water. Nylon loses roughly a tenth to a seventh of its strength wet, and recovers it on drying. Polyester and the high-modulus fibres barely notice.
  • Age and light. Ultraviolet is the quiet one. A polypropylene line left outdoors for a season can lose a large fraction of its strength while still looking serviceable.

Four

The fibre decides more than the knot does

Every knot on this page assumes a rope with some grip and some give. Modern high-modulus lines have neither, and knots that have held for centuries slip straight out of them. Before choosing a knot, know what is in your hand.

Typical properties of common rope fibres. Manufacturers’ figures vary; use these for choosing, not for calculating.
FibreFloatsStretch at breakWetSunlightMeltsKnotsWhere it belongs
Nylon (polyamide)no15-45%loses ~10-15%, recovers drypoor215-260 °Chold wellanchor rode, climbing rope, anywhere shock has to be absorbed
Polyesterno10-15%unaffectedgood250-260 °Chold wellhalyards, sheets, static lines, anything that must not stretch
Polypropyleneyes15-25%unaffectedvery poor160-170 °Cslip somewhatthrow lines, cheap utility cord, anything that must float
HMPE (Dyneema, Spectra)yes3-4%unaffectedgood144-152 °Cslip badlysplice it, do not knot it; creeps under sustained load
Aramid (Kevlar, Technora)no2-4%unaffectedpoorchars, ~400 °Cslip, and fatigueheat-exposed rigging; poor at repeated bending
Manila and natural fibreno10-15%swells, jams, rotsmoderatecharshold very welltraditional rigging, decorative work, anything a knot must bite into
0%10%20%30%40%50%HMPE (Dyneema, Spectra)3-4%Aramid (Kevlar, Technora)2-4%Polyester10-15%Polypropylene15-25%Nylon, static construction15-25%Nylon, dynamic climbing rope30-45%Manila10-15%
Elongation at break, by fibre. The spread is the point. A dynamic climbing rope is engineered to stretch a third of its length before it parts, because that stretch is what turns a fall into a survivable deceleration; a high-modulus line does not stretch at all, which is why it transmits every shock straight to whatever is holding it. Choosing rope by breaking strength alone ignores the axis that actually decides whether a system is safe.

Why knots slip out of high-modulus line

HMPE fibre is slippery by design. The same low friction that makes it run sweetly through blocks means a knot has almost nothing to bite on. Published tests routinely show knots in HMPE at half the efficiency they reach in polyester, and some knots that hold perfectly in nylon simply capsize and run. The fibre also creeps: under a sustained load it slowly, permanently elongates, so a knot that was tight last season can be loose now. The rope is spliced rather than knotted for both reasons, and a splice in HMPE keeps most of its strength.

Five

Which knot, in one table

Most rope problems are one of these. If yours is not, the honest answer is to work out which of the four families it belongs to and start there.

What you are doingKnotBecause
Tying a loaded line to a ring, rail or bollardRound turn and two half hitchesthe round turn takes the load before you tie anything, and lets you ease it out again
Making a loop you will need to untie afterwardsBowlineit releases after loading; back up the tail
Making a loop that will be loaded hard and inspectedFigure-eight on a bightstrong, and unmistakably right or wrong at a glance
A loop in the middle of a rope, no ends freeAlpine butterflytakes a three-way pull and does not jam
Joining two ropes of different diameterDouble sheet bendthe only common bend that tolerates a mismatch
Joining two ropes for good, over edgesDouble fisherman’scompact and symmetrical; expect to cut it off
Tensioning a load down onto a rack or trailerTrucker’s hitchrope-on-rope purchase, roughly 2:1 once friction is paid
A guy line you adjust by handMidshipman’s hitchslides when pushed, grips when pulled
Gripping a rope with a cordPrusikgrips loaded, slides free; watch cord diameter and heat
Temporarily hanging something off a postClove hitch plus a half hitchfast and tieable in a bight, but it slips on its own
Binding a bundle, or a rope end before whippingConstrictorit will not let go; tie it where you can cut it
Stopping an end running through a blockFigure eightbulky enough to stop, and it comes undone afterwards
Anything holding a person off the groundnone of the above, on their ownthat is a rated system and a taught skill, not a knot chosen off a page

Six

Ends, coils, and knowing when to stop using a rope

Most rope is thrown away long after it should have been, and most rope ends are finished in the way that lasts least.

Heat-sealed quick, brittle, and it fails when the bead cracks off Common whipping twine, whipped against the lay for a length of about one rope diameter, ends buried Alternating coil every second turn taken with the twist reversed, or the coil stores its own kinks
A melted end is a bead of plastic sitting on top of fibres that are no longer held together; when it cracks off, and it does, the rope unlays from there. A whipping is twine wrapped against the lay for about one rope diameter with both ends buried under the turns, and it outlives the rope. On natural fibre and three-strand rope, a back splice does the same job without twine.

When to retire a rope

  • Glazing or a hard, shiny patch. That is heat damage, usually from a rope running over rope or over an edge. The fibre underneath has already melted and reset.
  • Core visible through the sheath on a kernmantle rope, or a soft, mushy spot where the core has broken while the sheath still looks fine.
  • A shock load you can name. A fall arrested, a load dropped, a snatch on a tow line. The rope may have absorbed most of its remaining working life in one event.
  • Chemical contact. Acids destroy nylon and alkalis attack polyester, and neither leaves a mark you can see. A rope that has met an unknown chemical is finished.
  • Sunlight and time. Even stored properly, synthetic rope ages. Manufacturers commonly put an outer limit of about ten years from manufacture on life-safety rope regardless of use, and far less for rope that lives outdoors.

Keeping it usable

  • Coil so it does not learn a twist. Every turn taken the same way puts a twist into the rope; alternate the direction, or flake it into a bag instead. A rope that hockles (throws small loops when slack) has been coiled badly for years.
  • Wash it in fresh water if it has been in salt or grit. Salt crystals and sand are abrasives working inside the braid every time the rope bends.
  • Dry it out of the sun, loosely, not in a heap and not on a radiator.
  • Do not stand in the bight. Any loop of a rope under tension is a place where a broken end will travel. This is a habit, not a rule you remember once.
  • Log the hard loads. A retirement decision made from memory is a retirement decision made too late.

This page is a reference, not instruction

Everything here is written for rope work where a failure costs you a load, a tarp, or an afternoon. Rope that holds a person, in climbing, rescue, arborist work, work at height, caving or rigging above people, is a different activity with rated equipment, inspection regimes, redundancy and taught technique behind it, and nothing on a web page substitutes for being shown by someone competent and then practising under supervision. If you are about to use one of these knots for the first time in a situation where it matters, that is the wrong first time.

Seven

About this page, and about this domain

Short version: one page, no money in it, one named author, and a domain with a history that is stated here rather than hidden.

What this is

A single-page rope reference, not a blog and not a site that grows a page a week. If there is nothing worth adding, nothing gets added. It is one page on purpose: the whole thing prints, and it works offline once loaded.

Who writes it

Written and maintained by Mikhaila Molina, contactable at hello@iotlinefair.com and on LinkedIn. There is no house byline and no anonymous editorial team.

What is being claimed here is modest and worth stating plainly: this is a compiled reference, checked against primary sources (knotting literature, fibre manufacturers’ own published data, and the break-test studies named in the strength section) rather than a claim of professional rigging expertise. Where the sources disagree, the page gives a range and says so. Where a figure would be safety-critical, the page tells you to get the number from the manufacturer of the rope in your hands instead.

How it is written, including the automation

Google asks publishers to be open about how automation is used, so: the prose here is drafted with the help of a large language model working from the sources above, and is then checked and edited by a named human before it goes up. The diagrams are hand-authored vector drawings rather than stock art, written as coordinates in the page’s own source, which is why they are sharp at any size and why no image files are loaded at all.

Nothing is published at scale. There is no CMS, no content queue, no scheduled publishing, and no plan to produce a fixed number of pages per anything.

How it is funded

It is not. No advertising, no affiliate links, no sponsored or guest posts, no paid placements, and no links sold, exchanged or accepted in exchange for anything. Nothing here is for sale. The page carries exactly one outbound link, to the author’s professional profile. If any of that changes it will be said here first.

The history of this domain, in full

The page is called The Working End and the domain is iotlinefair.com. Those do not match, and the reason is worth stating rather than leaving a visitor to wonder. This domain has had four occupants and only the last is us.

  • 2015 to 2019, IoT Line Fair. A student-run technology fair organised by an EESTEC committee at Istanbul Technical University. A real event with no connection to this page. We do not claim its name or its reputation, and this publication is deliberately not named after the domain for that reason.
  • 2022 to 2024, bulk spam. After that registration lapsed, a different holder filled the domain with machine-generated casino, betting, payday-loan, dating and essay-writing pages across nine languages; public archives hold more than a thousand such URLs. Every one of them returns 404 now.
  • August 2026, a first rebuild. The domain was registered fresh from the registry on 5 August 2026. The first site under this ownership published measured results about home servers. Those pages have been removed too, and their URLs also return 404.
  • September 2026, this page. A different subject entirely, and a single page rather than a site.

The consequence of that second period is still live: the domain carries a manual spam action in Google Search, applied one day after the current owner verified it. That is stated here rather than left for someone to discover. It has been appealed and the appeal may not succeed, and this page is written on the assumption that it will not. A page only worth publishing when it ranks was never worth publishing.

How it is served, and corrections

One pre-built static HTML file on Vercel. No server-side code, no database, no user-agent detection, and therefore no way to show a search engine anything different from what you are reading. No trackers, no analytics, no cookies, and no fonts, scripts or images loaded from anywhere else. The page requests nothing but itself.

Mistakes get fixed in place and the revision date below changes when they are. Corrections and disagreements, especially about the numbers, go to hello@iotlinefair.com and are welcome.