Load Testing

Tests (full spreadsheet of all test data - including multiple test pulls for each item

 Quick statistical preamble

1.  Very very roughly the 95% statistical confidence interval for single pulls breaking strength is +- 5% of the line rated strength, and the 95% confidence interval for the mean of the 10 pull tests is +- 1.6% (of rated strength). And  the knot breaking strengths scale up relatively accurately as the line strength increases.

2.  The "slipping loads" are much less repeatable/accurate than the breaking loads.  In fact with some ropes, some knots (like the sheet bend) will slip part of the time and hold/break the other part of the time. It would be better to consider the slips as 'high' or 'low' rather than as specific lb numbers.  The slipping loads will scale to: the circumference of the line x friction coefficient of the cover.  This means the slipping loads will be lower (as a percent of rated line strength) for bigger lines (because circumference is a function of radius while strength is a function of radius ^2).

3. The variability of the breaking point test results were quite tight and repeatable for all lines, but it is perhaps worth noting that the variability was 3 times higher for the Samson lines than the New England Ropes lines.  This suggests somewhat higher consistency in the NER product.

Dacron Double Braid

1. These lines are less sensitive to knot construction than conventional wisdom suggests

a. The average breaking strengths are higher than commonly thought

b. The difference between the knots is less than commonly thought

Average breaking strength as a % of rated strength

(10 tests of each knot for 3 different lines)

Bowline Figure 8 Loop Buntline Double Sheet Bend Double Fisherman
74% 73% 74% 61% 67%
samson LS samson xls

Just as example of how relatively insensitive Dacron line is to specific knot construction, I tested a regular bowline vs. an Eskimo Bowline (which more would consider a completely incorrect way to tie a bowline). In single pulls in 1/4" stayset: regular bowline = 1610lbs (69% tensile), Eskimo Bowline = 1540lbs (66% of tensile), and my confidence interval is about 5%, so as we expected, the regular bowline is probably slightly stronger, but not by very much and not with statistical confidence (with one pull).  In theory the Eskimo bowline compensates that very slightly lower straight line strength with greater hoop strength inside the loop - e.g. if the loop is pulled in two different directions or you tie it around a large object, the Eskimo may then be stronger.

All the Dacron test results were stable across several different sizes (the difference between the two is that New England Ropes tends to put a higher breaking strength on their lines):

New England Ropes Stayset - Dacron double braid Samson LS - Dacron double braid
NEW bowline  samsonbowlinesize

These tests were for dry line. But for Dacron wet vs. dry does not make a significant difference.   For example, for bowlines, I got a 3 pull average breaking strength of 1780lbs dry vs. 1720lbs wet, with a 54lb stdev.  So that's not a significant difference at the 95% level. That is consistent with the literature, which suggests that nylon looses 10-15% when wet, but that polyester is pretty much the same.

2. The sheet bend slipped in all three lines.  These are new lines, and more slippery than used ones, but it still suggests the sheet bend may not be very reliable (see tests of alternatives below in the "specialty knots" section).

3. Slight differences exist in the stretch characteristics of the three lines, with Stayset being lower stretch (also almost twice as expensive).

stretch vs load

4. The key result from a drop test was that under the same falling load, a Dyneema line broke (at a figure 8 knot) while similarly rated nylon and polyester lines and webbing did not break. This is because the Dyneema does not have much stretch and thus creates much higher shock loads (stopping the same force in a much shorter period of time) than the nylon and Dyneema. Remember this if 'upgrading' to low stretch line - it will create higher shock loads and may break hardware that was fine before.

The rest of the drop test data was not usable, except to tell me something I had been unable to get from the load cell mfg -  that my load cell 'transient capture rate" is about 30hz (.035 secs) and it cannot accurately capture any event faster than that. That rate allowed me to measure the drop on nylon accurately, but the Dacron and Dyneema shock loads were faster. I can now design shock tests with durations lower than that for accurate measurement.  Simple engineering calculations, based on the published stretch numbers suggest these should be the drop impacts and durations (with 50lbs falling 4'): (drop impact load calculator)

stretch vs strain 

It does provide a vivid reminder that moving to a 'high tech'/high modulus/low stretch line will dramatically increase shock loads on your gear.

5. With polyester line, a lubricating finish can significantly improve fatigue resistance.


Dyneema Single Brand

1. Few common knots will hold in these lines. Of the common ones, the Figure 8 was the only one that held.  The Bowline, Buntline, Sheet Bend, Double Sheet Bend and Double fisherman all slipped.

 The two 'easy' ways ( lightly sewing the tail down or tying an overhand knot in the tail) often used to try to stop knots slipping, will usually not work in Dyneema single braid.  When the knot slips, the tail will receive near full load, which will break anything less than full strength stitching and cause stopper knots in the tail to slip. The following pictures shows a double fisherman, with the tails sewn down, and then after being loaded up and started slipping the sewing easily ripped out. Note: it is possible to successfully sew down the tails, IF you first preload the double fisherman so that it is rock hard/tight and absolutely all the slack is out of the knot, and then you sew a bunch of stitching using doubled 80lb fishing line.  If you do this, the stitching will load up pretty highly but not break (if you put in enough) and the double fisherman will break at about 50% of the line tensile strength.

before after

2. In agreement with the conventional wisdom that these lines are particularly sensitive to knots, the Figure 8 only retained 42% of the line strength.  The Figure 9 knot retained 44%.  The Water Bowline is another (less common) fixed loop that will hold and not slip in bare Dyneema. The simple water Bowline is about the same strength as the Figure 8, but a modification where the tail is tucked thru the knot is stronger - around 50% of line strength.

3. There are fishing knots specifically designed for slippery line.  Of the ones tested so far, the Palomar Knot is the best, not slipping and retaining 54% of line strength, but it requires you take a loop completely over/around the shackle (which you can not do with say a mounted pad eye). While messing around with this, I accidentally invented a knot , which weare calling this the "EStar knot" after my forum username, that does not slip, is about as strong as the Palomar, and does not require "taking the loop around the shackle'', so it is more general purpose than the Palomar.  Some sailors have been using a simpler buntline modification, but this knot will slip in bare Dyneema, at around 35% of the line strength. And there is a third modified buntline used by sailors, where you simply tie an overhand knot in the tail and snug it up against the buntline, but it also slips in new bare Dyneema. The EStar knot is so far the best "easy to tie, no slip and high strength, general purpose knot" I have tested in bare Dyneema.

Sequence For "EStar" knot

(start with a buntline and then . . . )

bunt1 bunt2 bunt3

4. I tested 10 soft shackles at an average 170% of the underlying line strength.  The two keys to soft shackle strength are (a) ensuring the length/tension on the two lines in the shackle are exactly equal, and that the cover is milked down very hard and smoothly on the cover/bury portion of the shackle, and (b) reducing the loading on the diamond knot (which is where they always break).  This can be accomplished by using longer soft shackles of lighter cord, and looping them between the load points several times (like a lashing) before closing the diamond/noose.  Note: Soft Shackles 'theoretical' max strength is 4x the line used, because they are constructed with two strands in a loop (so four strands between the two load points).  The diamond knot then reduces the strength by 50-60%, so you end up at about 170-200% of line strength. Note: In some other people's testing, some soft shackles have tested only at around 100% of the line strength.  That will happen if, for example, only one strand of the shackle is carrying all the load. 100% seems to be the 'floor strength' and is a safe/conservative assumption, but a carefully constructed one will/can be significantly stronger than this.

Update: I just tested two soft shackles at 194% and 195% of the line strength, by being extremely careful to milk the cover very tightly and being extremely careful to get everything the same length/tension.  I also tested whether it made any difference where the diamond knot is with respect to the loaded pin, and it does not seem to make any significant difference. The below is the highest one I tested, but I suspect that has more to do with more careful construction than the diamond location.

diamond soft shackle

The soft shackles always break at the diamond knot, so I tested other stopper knot designs to see if we could strengthen that component.  As expected, the diamond breaks at 46% (of line strength), which is exactly consistent with tested 170-180% soft shackle strength (400% double loop strength x .46 = 184%, with perhaps 10% additional loss in the locking and bury and construction imperfections).  The Ashley stopper knot, which looks the business, slipped at 27%, and the double fisherman stopper slipped at 15%.  So the other common stoppers cannot be used in the soft shackle. The Palomar and EStar knot can be converted to stoppers which test at about 50%, but they do not have the nice 'ball shape' that the diamond has and I am not sure the extra 4% is worth giving up the good diamond shape (which contributes to the soft shackles never coming accidentally undone). 

 I have found that that pulling 'hand tight' knots like the diamond (e.g. knots with a bunch of bulk and size) very fast (but still within the cordage institute standard) would significantly decrease the breaking strength. The practical lesson for us is to 'pre-tension' our knots (on soft shackles and fixed loops) quite slowly to a very high load (say right to or even just past the expected working load limit), so there will not be much slippage left and heat if they are shock loaded.

It is possible/easy to replace the diamond knot weak point with a spliced on aluminum toggle (3/8" rod x 2").  This will increase the strength by about 30% vs. the diamond knot, and the soft shackle will then break at the small bend radius where the 'noose' goes around the line just below the toggle.


Improved soft shackle: The conventional soft shackles all break at the diamond knot.   So the question to explore was: is there an easy to tie replacement for the diamond that is stronger?  The answer turns out to be yes. By burying the tails into the shackle where they exit the stopper knot you can make a soft shackle which is 230% of line strength, which is about a 50% increase over the conventional diamond soft soft shackles, and which moves the weak point to the bend radius at the noose.  This can not be neatly done with the conventional diamond stopper knot, so three alternatives are described below. All three are the same strength. Each of the three has pro's and con's.

(a) The 'overhand/loop' stopper. This is easy to form, but requires a relatively high load to tighten up properly.  You will probably have to tension it up with a winch to tighten the stopper knot properly.  This solution is explained in detail here, with a quick summary below:

essoft3 es1w es4w

(b) The 'loop' stopper.  This is even easier than the above, can be tightened by hand, and makes a smaller stopper knot, but it requires that the eyes be formed at the right size - just big enough to allow the noose end to be pushed thru and no bigger.  If the eyes are made too big the knot will turn inside out under load.  It is not hard to make the eyes properly, it just requires careful attention to that step.

Form two eye's in the tails, make the eyes just the size of two
pieces of the line.  Then push the 'noose end' of the soft shackle
thru both eyes.
Then run the noose end thru the bigger loop you have
 just formed
Tighten up the knot
loops soft 1 loop soft 2 loop soft 3

(c) The 'Button' stopper (credit to Brion Toss).  This is a complex knot to tie, even more complex than the diamond.  But it forms a very attractive round symmetrical stopper knot, which can be tightened by hand.  The two key things to pay attention to with this solution are making sure the knot is properly constructed (in the later stages it is easy to get a tuck in the wrong place) and making sure the final tail buries are both long enough and started as close as possible to the knot body. Instructions.

Form the normal 'noose' at the mid-point of the line From the button knot with the two tails.
Instructions here.
The tails of the button extend down along side the shackle parts.
Bury the tails into the shackle.
noose button bury

And if you want a 'simplier soft shackle' in the end of a line - for say a halyard or jib sheets . . . here is one that is 99% of line strength and very easy to make: simplier soft shackle instructions

First you bury a 'reinforcing' piece of line in the center
 of the primary line
Then you form the 'noose' thru the center of the
reinforced section
Finally you tie a stopper knot on the end.  This is
an estar stopper.
thread push thru completed

By the way, it is also perfectly possible to make soft shackles in dacron line.  It is easiest in 12 strand hollow braid, like New England Ropes Regatta Braid and Samson's Tenex.  Below is a soft shackle made in 5/16" Regatta Braid - it broke at 6010lbs (at a tight bend at a pin end), which is sufficient strong for many applications and it has a very nice feel/hand to it.  This is consistent with our theory that this 'should be' about 220-230% strength (of the line tensile) given bigger diameter puller pins.  During construction, carefully getting the two legs the same length is important with the dacron soft shackles; because the dacron has much more friction, so the diamond knot does not tighten up as much as in dyneema, which does not allow the legs to 'equalize length' as much as the shackle is loaded up for the first time.

dacronsoft1 dacronsoft2 

5. Tapered straight bury splices are potentially just a little stronger than the Brummel splice.  The key to high strength in the straight bury is to finely and carefully taper the buried tail. The splice will usually break at the end of the bury unless it is very carefully tapered. If you do a 'bad' splice, with the end buried but no taper at all in the bury, the strength will be about 86%. If you do an even worse splice, with the end sticking out and not buried at all, it will be 80%.

The optimal taper length appears to be around 32 - 48 diameters.


I have just done a careful test of 7 samples, with loops on each end, all with 64 diameter bury's, all with the same size eyes. They were pulled with flat pins with 4:1 bend radius.  The only variable was taper length.

 - Sample #1 had no taper, just the line cut off square.  This broke at 84% of line tensile, which is exactly consistent with my previous test where it broke at 85%. Note also in that prior series, a 'whoopee sling' exit (where the tail is left hanging out of the bury) broke at 80%.

 - Sample 2 had a smooth (e.g. strands cut away at equal intervals)  8 diameter taper.  This broke at 91%.

 - Sample 3 had a smooth 16 diameter taper.  This broke at 95%

 - Sample 4 had a smooth 32 diameter taper.  This broke at 100%

 - Sample 5 has a smooth 48 diameter taper.  This broke at 99% - statistically the same as the 32 diameter one.

 The above all broke some where between the end of the taper and the tapered line - it is difficult to determine exactly where, but not in the bury and not in the loop.

 - Sample 6 had a 'accelerate' 48 diameter taper.  In this half the strands were cut away in the first 8 diameters, and the other half in the remaining length.  This broke at 90%, in the area of the 'accelerated' taper- obviously the 'acceleration' was too severe.  This is statistically the same as sample #2. 

- Sample 7, the "Samson taper", a taper at 1 fid (21 diameters) from the end, constructed by cutting every other strand pair until half the strands are cut. Then cut the remaining strands at a sharp angle at the end to that it is very pointy gradual taper but not strand by strand, just cut all the strands at an angle.  This tests at 100%.  It is slightly more length efficient than the # 4 & 5 'smooth' tapers'.

 This all leads me to believe we have an asymptotic (to 100%) curve, with little meaningful gain past a 32 diameter bury 

The single Brummel (alone, without any bury) breaks at about (or just a little less than) 50% of line strength.  If it is just a little less, then the brummel with bury splice will be a little weaker than the straight bury splice.  But in my test, the 'or just a little less' is within the statistical variability and I can not prove it either way.  A double brummel is 66% of line strength, and a five brummel is 78%.  None of them slip, which is the main advantage of putting a brummel into the splice - it will not slip out when flogging at low/no load, as is possible with an unsewn straight bury (but that can be cured just by putting a couple stitches thru it). Here are instructions for an alternative way to 'lock' the splice, used by commercial heavy lifting cordage providers

Regarding bury length . . . quite short bury's will carry full strength.  Under a slow 'static load' 18:1 will slip, but 27:1 will hold.  I tested 27:1, 37:1 and 46:1 bury's under a significant drop load (50lbs falling on a 4' 1/2" dia Dyneema line as the 'fall line').  All three samples slipped about 3 diameters, which I take to be the "constructional take up" of the splice (without any stitching), but then all three held.  Extra bury length is of course always a nice thing and double 27:1 = 54:1 (which is near the typical bury recommendation) would seem to be conservative, but these results do suggest you can 'get away' with a shorter bury if it is necessary to make the piece.

When you do a bury splice, the cover will 'shrink in length'. The cover expands in diameter and contracts in length as you put the core inside.  It shrinks 20% (19.3% in my test samples) of the length of the bury. So, for example on a 5mm line, with a 30cm bury (60:1), it will shrink 5.8cm.  But when you load up the line, both the splice and the line itself will 'grow, due to permanent constructional stretch. The splices will grow 10% (3.25cm in the 5mm line samples) and the line itself will grow 3.4%.  Thus, the splices will 'grow' (due to constructional stretch) about half the amount they shrunk (due to bury), leaving a net shrinkage of about 10% (2.55cm in our 5mm line example).

 But the line between the splices will grow 3.4%.  This exactly offsets the net splice shrinkage if the line between the splices is 150cm long.  There will be net growth if the line is longer.

6. A sewn splice retained 81% of the line strength, before breaking at the first set of stitches.

7. The stretch characteristics of the two tested lines (Amsteel Blue and Endura12) were (statistically) identical within my ability to measure.

8. UV exposure damages all lines and fibers, but it particularly degrades the strength of single braids, as there is no cover to shield the load bearing core fibers.  How much the line strength will be degraded depends on what sort of coatings are applied to the fibers and exactly what sort of braid is used.  The following graph, based on tests conducted by manufacturers, shows the high and low cases for UV damage to Dyneema single braid lines.  Most Dyneema single braids will fall somewhere between these two lines.

uv damage

Dyneema Double Braid (Dyneema core and Dacron cover)

This rope construction is generally used for two reasons: (a) to protect the core from UV, and (b) to make the rope less slippery in clutches and knots. However, for (b) the Dacron cover appears to be a limiting factor.  The cover tears at the knot, and then the Dyneema core slides thru. Note: these cover breaking strengths were pretty much identical in two different Mfgs Dyneema double braid line (Samson WarpSpeed, and RobLine Dinghy Control line)

Bowline EStar Knot Halyard Knot Buntline
Cover tore at 52% of line strength Cover Tore at 62% of line strength Cover Tore at 48% of line strength Cover Tore at 42% of line strength
bowline fse Estar knot halyard buntline fse

1. So far, the plain old bowline looks like a pretty good choice for a fixed loop (and the EStar for a sliding loop) for these high modulus covered lines. . . . very unlike how bad a choice the bowline is for the high modulus single braids. But, as with the Dyneema single braids, splices are the best answer.

2. There are some specialty Dyneema double braid with Dyneema (rather than Dacron) covers (like NER's WR2 line).  This tends to perform like a Dyneema single braid of the same diameter at the core - e.g. the cover provides chafe and uv protection but not added strength nor added knot holding power.  A bowline will slip at about 32% (of line strength) until the bowline jams up against the pin, and then the cover will break and the core slide thru the knot at about 47%.  The EStar knot will break (both core and cover together) at 57%. Here is a picture of the bowline tight against the pin and the cover broken:


Note: I took some of the NEW WR2 product apart and tested the core and cover separately (using pieces with careful splices on each end).  The core broke at average 1210lbs and the cover at 1585lbs.  This line is rated at 1800lbs, and NER's splicing instructions is to strip the cover off and do a core to core splice exactly as I did in the core samples. I did a sample with the complete rope (core + cover) and it broke at 1410lbs (17% more than the core alone), so the cover is adding some strength (but it was still 20% below rated strength) and that has implications about how to splice it  You should not just do a core splice and then slide the cover back up and over the taper - you should fully engage the cover in the splice.

3. The cover may not protect the core as much as one (or I) would expect.  It is hard to achieve repeatable and consistent results when testing old line,  but testing some 15 year old line suggested the core was only (roughly) 40% of original strength.

4. There is some test data indicating that UV will penetrate 4mm thru/into the typical braided cover.  So if the cover is les than 4mm thick, as it is on most yachting lines (my 9/16" double braid has roughly a 3mm thick cover), then the core will be damaged by exposure to the UV.

3 Strand Nylon

1. Similar to the Dacron Double Braid, this low modulus line is less sensitive to knots than commonly thought.

Average breaking strength as a % of rated strength

(10 tests of each knot)

Bowline Figure 8 Loop Buntline Sheet Bend Double Fisherman
70% 73% 86% 59% 74%

These were tested in dry line.  Wet nylon is a little weaker.  The industry says it is 10-15% weaker. With 3 bowline pulls, I found a dry breaking strength = 1190lbs and wet=1107lbs, with a stedev of 56 and a 95% CI of +- 64, so that is a statistically significant result (7% less when wet in this case).

2. As expected, Nylon Double Braid is the least stretchy.  But the difference between good old Three-Strand and Brait was not as big as expected, and at least with these particular lines (Yale 1/2" Brait and Samson Pro-set 1/2") and loads the Three-Strand was marginally stretchier.

nylon stretch

3. I have conducted some 'drop test' to measure shock absorption. . . . 50lbs dropped from 46", the data from the second test on each line:

drop test 

The thing I find most interesting is how much lower the shock load is with 1/4" than 1/2" line.  I know it 'should be' because the smaller line is at a higher % of its tensile strength and thus will stretch more, but it is interesting to see first hand the 50% more load on the 1/2" than the 1/4" 3 strand. This has implications for anchor chain snubbers and tethers.

The rest is pretty much as expected.  Brait has the most energy absorption, 3 strand next and double braid least.

Dynamic Climbing Line

The is specifically designed to be (a) elastic to cushion a climbing fall, while (b) having a chafe resistant cover.

1. It is astonishing elastic.  Just for reference, nylon braid, which is the most elastic 'marine' line will stretch 13% at 30% of its breaking strength.  This dynamic line stretched 35-40% (for the first heavy loading).

dynamic line stretch

2. Some of the above stretch is 'permanent constructional elongation' (e.g. the rope becomes permanently longer and thinner), which occurs during the first few heavy loadings. And it is no longer available for shock absorbing after that.

constructional elongation

So after 5 pulls to 30% of breaking strength, while under load, the rope will be 'stretched' 35%; and 15% of that will be 'permanent' and not shock absorbing any more, while 20% will still be elastic and shock absorbing. Note this 20% is still 50% more shock absorbing capability than the brait.

This makes it really attractive for areas where you want shock absorbing - chain snubbers, traveler lines (to soak up jybe shocks) and tethers.  We have used it for 20 years as a snubber and traveler line, and its performance supports both its shock absorbing capability and its durability.

3. For reference I tested a west marine tether, using nylon webbing.

  7.7mm Climbing line West Marine Webbing
Permanent elongation after 5 pulls 15% 5%
Elasticity at 800lbs load 20% 10%
Elasticity at 30% of breaking strength 20% 15%
Impact force on body in ISO drop test 1600lbs 2700lbs
% of dynamic breaking strength in ISO drop test 90% 77%

So the west marine tether will have a 68% higher impact force on the body (probably less than this in the real world due to clothing and fat compression); but also has a higher 'breaking strength safety factor'.  If we take the ISO test case as a 'worst case', then the 7.7mm climbing line solution is preferred. If more 'safety factor' is desired then moving to 8.5mm climbing line would probably be preferable. Note: none of this includes possible elasticity from knots in the climbing line, which would make it even 'softer' impact load and increase its safety factor

4. It is not easy to splice (because the cover is very tight braided) and a splice will meaningfully diminish the shock absorbing characteristic.  Climbers use figure 8 knots - which in of themselves add even more shock absorbing as they tighten up.  But my feeling is that is not ideal and sewn splices may be better for the typical sailing applications.  Properly done sewn splices will be both stronger (useful for the snubber) and more compact (useful for the tether).

Test data for 7.7mm dynamic line:

  Breaking load % of line strength
Line strength 2671lbs 100%
Sewn Splice 2270Lbs 85%
Figure 8 loop 1900lbs 71%
Overhand loop 1640lbs 61%



Sewn splice Figure 8 loop  
sewn1 eight1  
sewn2 eight2  

Here is an interesting test of climbing tethers.  Their conclusion is that sewn webbing tethers are unacceptable because they do not have much shock absorbing capability, and can easily impose body breaking shock loads.  They recommend knotted dynamic line tethers. Interestingly, their tests suggest that even poorly tied knots are better than sewn terminations.

Specialty Knots

Gripper knots

1. Grip test - Pulling on a slippery 1/2" Dyneema single braid line, with 3/16" Dacron Double Braid gripper line (Samson LS - rated at 1200lbs)

- Rolling hitch slipped at 200lbs

- Rolling hitch with 3 turns slipped at 370 lbs

- Prusik slipped at 860 lbs

- Icicle hitch did not slip, Dacron gripper line broke at 900lbs (75% of rated strength)

2. Strength test - Pulling on 1/4" chain with 1/4" Dacron double braid (stayset) loop (tied with a double fisherman).  The first thing I discovered was that you need a minimum of a 3.5:1 bend radius at the snubber attachment end of the gripper loop, or it will break at that bend rather than at the chain.

- Rolling hitch broke at 2150lbs (50% of loop tensile)

- Prusik broke at 3240 lbs (69% of loop tensile - statistically the same as the Klemheist below)

- Klemheist broke at 3280lbs (70% of loop tensile)

The Rolling hitch was weakest, and had least grip, and I personally find hardest to undue after string loading.  So I would scratch it from the snubber system.  Between the Prusik and Klemheist it's a toss up, but I personally find the Klemhesit a little easier to undo after strong loading.  The Icicle is the choice for very slippery line.

Jib Sheet Attachment

Some people use one long continuous jib sheet and attach it in the middle to the jib clew, using either a Luggage Tag or a Clove Hitch.  In Dacron Double Braid, the Luggage Tag and the Clove Hitch will both slip slowly, the Luggage Tag at about 40% of line strength, and the Clove hitch at around 60% of line strength. By comparison, the Bowline broke at 70%. After high loading, the Luggage Tag can be undone using a spike, while the Clove Hitch usually has to be cut off with a knife. A  Bowline will break before the Prusik Hitch slips, and both can be untied (with a spike). So, there is a decision to be made about what failure mode you prefer . . . . . potentially a slow slip (luggage and Clove) or higher strength but potentially a "Bang" (Prusik) and how important it is to be able to get the sheet off the jib intact (Luggage & Prusik).

Luggage Tag Clove Hitch Prusik Hitch
(Luggage tag with additional turn)
Slips at 41% (of line rated strength) Slips at 59% Did not slip
Broke at bowline
cow hitch clove hitch prusik 



The sheet bend is perhaps the most common knot to join two lines.  But it slips (very roughly) 50% of the time in brand new Dacron line and breaks at about 48% of the rated line strength (which is low among the various good Dacron knots).  The Zeppelin bend is an equally easy knot to tie, does not slip and breaks at 58% of rated line strength. It seems a better choice for 'everyday' use. Note: the bends in this section were pulled tested as "loops". I have measured and adjusted the results for loop friction at the pins.  But the results will be somewhat less accurate/reliable than the 'straight pulls' done elsewhere on this page. 


Sheet Bend Zeppelin Bend Strait Bend
Slips about 50% of the time in brand new Dacron
Breaks at 48% of rated strength - Dacron
Slips in bare Dyneema
Breaks at 58% - Dacron
Slips in bare Dyneema
Breaks at 48% - Dacron
Slips in bare Dyneema
sheet zeppelin strait

Bends are very difficult in Dyneema single braid - almost all slip at relatively low load, even the much touted triple fisherman.  My best solution right now is to tie two EStar loops together.  This will be secure, strong and low profile. All the other bends we have tested have slipped or been complex to tie or both, and none stronger.

An alternative to both a bend and 'back to back' loops is to use the eye structures (of the non-slipping eyes), but make the first part of whatever eye (say, Fig.8) on each line and then the knots are completed with each other's tails (end-A with end-B's, ...).  That will equal the non-slip performance and strength of the eyes alone, but with out the chafe and 1:1 bend radius of interlocked loops.  That's a bit complicated in writing, but perhaps a photo will help.  These are figure 8 'loops', done in the above fashion:

Figure 8

Water bowline

eight1 bend water1 bend
eight2 bend water2 bend
eight3 bend water3 bend

And tested in Dyneema they show the non-slip performance we hoped for (this is the figure 8 loop with two of these joins shown/tested, and broken in the right one):

eight1 web bend eight2 web bend
Fishing Knots

All these tests in 7/64" Amsteel (single pulls, except the two knots which slipped, where I did three just to confirm).

 Two old reliable's as benchmarks:   Figure 8 = broke at 900lbs and Water Bowline = broke at 900lbs

 - EStar = broke at 960lbs (given the single pull that is on the edge of significant/not significant stronger than the above two)

 - Orvis = slipped at 790lbs

 - Uni = slipped at 973lbs (range from 860 - 1110lbs)

 - Trilene = broke at 1090lbs

I was surprised when the two slipped.  I always thought of fishing line as the 'most slippery', but this Amsteel with polished stainless pins appears to be even more slippery.   I am truly skeptical of anyone claiming knots 'holding near line strength' in Amsteel. The only ' near knot like structure' that I have tested that clearly scores well beyond 50% in Amsteel is the multiple (5x) brummel (without bury) - I tested it at 80%. But it is really more like a 3 strand splice than a true knot.

Rope to chain splices

There are three popular rope-chain splicing methods.  All are 'full strength' (if done properly).

The three I tested are: (a) regular "three-strand loop splice", ( b ) the "two strands through the chain link in one direction and the third strand the other way, loop splice", and ( c ) the "up the chain weave."  All broke at 'full line strength.'  At the 'bend radius' around the chain link, the 3 strands separate a bit and mash down in all three splices and the bend vs. any one strand is more than 1:1.

So, the answer here seems to be "it does not matter so long as you do any of the splices well".

Dyneema Lashings

I tested 5 Dyneema lashings, in three sizes of line (these were all '3 round turn' lashings). I was curious how resistant the half hitches are to slipping, and where they would break. The failure mode was highly variable but the failure load was roughly the same (about 50% of 6 x line strength).  After doing a few more samples with different configuration I concluded that the key weakness is that the different loops in the lashing do not full equalize the load, in part because they are clamped by the half hitches.

 I then tried a couple samples where I half hitched only around the last strand, rather than around the whole lashing.  This was more prone to slipping and required 5 half hitches to be secure.  But it also had a 30% higher average breaking strength (probably because it allows more equalization of strand load).  So, it world seem to be a "better" solution without extra work/cost.  These did not break at the half hitches, so a 'better' constrictor knot than the half hitch would not appear to add any strength (but might add extra slip resistance).

Lashing produce less load multiplier than normally expected, because of rope to rope friction at the bend points. I set-up a 6:1 lashing between two 1/2" carabineers (similar bend radius to a thimble that might be used in a life line) using 1/8" Amsteel - I put 50lbs on the tail (I did that by hydraulics so I could hold it steady, but that is about what you can pull by hand) and played with the lashing a bit to try to equalize tension between the strands - there is a ton of friction and binding as they go around the carabineers - and I got a steady state 100lbs on the tackle.  That is way lower than I expected - only 2x the tail load rather than the 4 or 5x most people expect.  I got a momentary peak loading, as I was plucking/equalizing the lashing strands, of 200lbs. To remove the 'inter rope' friction, I then set up the 6:1 lashing with each turn on a separate carabineer.  I still only got a 110lb total lashing/tackle pull - 10% up from having all the rope together on a carabineer. This suggests the vast majority of the friction is rope on bend, and the rope on rope effect is relatively small. I am really surprised at how much friction and how little multiplier effect there is. So the summary is:

2:1 multiplier when all loops thru same two solid bends
 (e.g. no sheave, 4:1 bend ratio)
2.2:1 multiplier when loops thru separate bends
 to eliminate rope on rope friction
4.4:1 multiplier when thru small ball bearing blocks
lash1 lash2 lash3

If you need adjustable leverage/purchase with the low friction rings the best way is with a cascade.  The below 8:1 cascade produced 260lbs force with a 50lb pull, or 5.2:1 leverage.  That is 65% "efficiency" (5.2/8). Which is much better than the above straight lashings at 33% (2/6), but not quite as good as the small blocks at 73%. An 'all Dyneema' cascade, replacing the carabineers with loops, generates 45% efficiency, still better than the lashings.

8:1 Cascade
with 4:1 Bend radius = 65% efficiency (260lbs output)
 All Dyneema 8:1 Cascade
with 1:1 bend radius = 45% efficiency (180lbs output)
(still better than the lashings)
cascade cascade2

Sometimes you want to make a very strong sling from a large number of light loops.  The big question with such constructions is how to join the ends - so that they are low profile but still non-slip and strong enough.

The first important observation is that 'strong enough' does not have to be very strong. On a 5 loop sling (so 10 'legs' holding the load) pulled to 1000lbs the tails are loaded only at 40lbs. And for a 10 loop sling (so 20 'legs' holding the load), the tails are only loaded at 7lbs.

5 loops of 1/8" amsteel, loaded to 1000lbs,
generated 40lbs load on the tails
10 loops of 80lbs dyneema fishing line loaded to 1000lbs,
generated 7lbs load on the tails
5 loop 10 loop

So, the question is how best (easy, low profile) to joining the ends to hold those sorts of small loads.  I tested four methods.  Sewing the tails ends together seems like the best solution - strongest and low profile for +7/64" line.  To make very high count loops, using smaller line (like Dyneema fishing line), which are harder to sew, one of the good bends (perhaps double fisherman) is the best solution.

A very simple low profile method is to just butt the two ends
together with adhesive lines heat shrink.  The tubing stretches
apart at 65lbs.
A second method which is slightly more work, but also very low profile
is to sew the bitts together.  This is with 80lbs Dyneema fishing line.  tying
off the fishing line with two half hitches and a square knot slipped at 130lbs.
Tying the fishing line off with a zeppelin bend pulled the amsteel braid apart
at 195lbs.
heat shrink butt joint sewn butt joint
A slightly bigger profile technique is an overlapped joint with
adhesive heat shrink tubing.  Here one end of the line pulled
out at 155lbs.
And an overlapped joint with 3M electrical tape, pulled apart at 50lbs
heat shrink overlap tape overlap
Stopper knots

There are various stopper knots that all work in Dacron - Figure 8, Stevedore, double overhand, and Ashley; but they all slip in Dyneema single braid.  Taking the EStar, one of the few sliding loop knots that does not slip in Dyneema single braid, and sticking the tail thru the loop before tightening it up, produces a 'better stopper knot', which will not slip and is nicely shaped.

Low Friction Rings

The 'best' solution is usually a splice (see caution about throat angle in the 'bend' section below), but sometime you want to be able to tie them in place (as when using as a movable snatch block), and sometimes you want to use the outside of the rig for the working line rather than the more typical inside application.  To do this we need a tie that will close/clamp the outside of the ring, so the working line will not slip out, but will also center the ring and hold it symmetrically (which will not happen if you tie most of the sliding loops, or luggage tag or prussik a loop).

The solution here is a clove hitch (with the clove hitch "cross over" inside the ring), and then a water bowline.  The clove hitch will clamp, while the water bowline will center it.  The water bowline is no-slip in bare Dyneema.

low friction ring


Hardware for making loops without splices

There are a number of products sold which are designed to make loops without the difficulty of splices.  The question is: are they any better than a free knot.  The answer is NO.

The EZ Splice is a piece of plastic with two holes drilled in it and a dozen stainless steel nails.  You stick the rope thru the holes and then nail it in place.  It actually looks and feels pretty impressive, but in 1/2" nylon line it broke at 36% of line tensile, when a bowline would go to 70%.

ez1 ez2
ez3 ez4

The Clamp-it is a tool for wrapping wire tightly and neatly around things.  It is primarily designed as a hose clamp replacement. The mfg does not make much of its potential for making loops in rope but does mention it in passing.  The problem with these wire clamps in nylon, is that as the rope loads up it gets thinner, and thus slips out from the clamps, at 12% of 1/2" nylon tensile.  In the picture at right below, the loop has collapsed, and the tail now starts slipping out  I have tested lower stretch line (stayset X)-see a bit below.

ct1 ct2

Stayset X is a low stretch parallel core dacron line.  It will not get as thin under load as nylon will.  I also tried a slightly more sophisticated whipping design.  These two things improved the results by 50% (1000lbs to 1500lbs) from the above nylon results, but it is still way way less than a bowline.

wire dacron

Velcro webbing loops

A piece of webbing, with velcro sewn on both sides, is commonly used as a mainsail clew strap.  These are surprisingly strong, and there may be other possible applications.

This is "genuine Velcro brand, made in USA", I bought at WM, which is not a 3m product, and appears to be weaker than the 3m product.  So, first lesson is that there are differences in strength between the various hook and loop products.   I pulled between two shackles.

 - A straight line shear pull (eg no loops) I got 7.5 psi.

 - A "2-wrap" I got 45 psi (6X the straight line - I 'expected' a bit more than 4x, and got +2x more)

 - A "3 wrap" I got 70 psi (9.3x the straight line - I 'expected' a bit more than 6x, and got +3.3x more)

 So, it would appear there is a lot of advantage to the first loop, probably because it puts a compression friction loading on the velcro, and then significant but diminishing returns to further loops.   Extrapolating a bit - A 6' webbing piece made into a 3-wrap loop, of the 3m tape, should then be 72 sq in x 14psi x 9.3 =  9400lbs (if the 14psi spec is correct - I have some on order to test/confirm).   If I have this right . . . . That's pretty damn strong, dyneema soft shackle country - there may be an opportunity to use these in more applications. The failure mode is also nice . . . a very slow creep, rather than just letting go.

The 'sewing theory' is that zig zag is used when there was going to be a sideways load (to the stitch line), but that zigzag decreased velcro hook and loop bond strength. However I could find no test results to support that theory. So, I created three samples.  All with small puller loops sewn on the ends and 5 sq in of velcro to velcro bond overlap area.  One sample had no stitching at all in the overlap area - simulating an adhesive solution.  One solution had two rows of short straight stitching. And one sample had two rows of long zig zag.  I pulled them until they came apart.

The sample with no stitching in the bond area was by far the strongest (35% stronger than the straight stitching), and in fact it was the only sample where the velcro fabric broke at the loop rather than the velcro bind pulling apart. I tested two different versions of this sample (one with better sewing at the loop) to be sure and both broke the fabric rather than pulling apart the velcro to velcro bond. Straight stitching was 10% stronger than the zig zag, both having the velcro to velcro bond pulling apart.

Message - for strongest bond strength, use adhesive and the least possible sewing in the bond area, and short tight straight stitching rather than zig zag.  There are two difficulties with adhesives alone.  The first is that webbing/fabric is not a great substrait for most adhesives (but as tested below in the webbing section, for example, 3M spray 90 properly applied, can generate 100psi bond to webbing) , and the second is that the 'hook side' of velcro is stiff when bending and without any stitching at all will 'gap' off the base webbing/fabric if there is a bend.  So, the most practical solution is adhesive plus just  enough stitching (straight short tight stitches right near the edges of the velcro) to clamp the two surfaces together.

velcro sewing

Bend radius


1. Bends reduce both static line strength and fatigue life, per the following graphs. A 1:1 bend reduces line strength 50%, and a 4:1 bend reduces line strength 25%. A close approximation to this "breaking strenght loss due to bending" is = 1 - .5 / (D/d)^.5.

bend strength Bend fatigue


2. There are three quite different cases to consider:

(a) A loop (spliced or knotted) over a pin.  The key factor to realize in this case is that the spliced loop has two static legs and thus the load on each is half the load on the main line before the splice.  So, the splice is starting at 200% of the main line strength.  Introducing 1:1 bend will cut that in half, leaving it at 100%, so the entire system is equally strong. This is confirmed by my testing - Dyneema single braid splices, joined 'loop to loop' (e.g. with a 1:1 bend radius) broke at an average 98% of rated line strength, and did not break at the bend but rather at the splice tail taper end.

 If there is a knot in the system it will break at the knot. If there is no knot and the splice is not precisely made the system will break at the splice. If the splice is perfect then the line could break anywhere. But the main point is that a 1:1 bend radius is acceptable inside a static loop.  An additional important point is that the ratio of the length of the eye splice to the diameter of the object over which it is placed would be a minimum of 3:1 and preferably 5:1.

bend radius 2 peeling load

What impact these 'peeling loads' with have on splice strength will depend in part on the specific splice and line.  But to gauge rough impact on strength I built three samples with 1:1 bend radius and 3 with 2:1 radius. The 2:1 bend radius samples all broke at the splice taper end near the line rated strength.  While the 1:1 bend radius lines broke at the throat at an average of 75% of line strength (note: one of these pulls broke at what seemed an anomalous low value, but I left it in the average because I could not see anything 'wrong' with it).  So, for this splice (simple bury) I would say that 1:1 loses some (15-20%) strength but 2:1 is near full strength.


One option get both a long throat angle and still 'clamp' the low friction rig, is to first make a constrictor splice around the ring, before then making the bury splice further down the line with an appropriate throat angle.  In the picture below, the constrictor splice is made, and the bury splice about to be made at the appropriate throat angle.


(b) A static, end loaded, line on a pin (like a mooring line on a bollard, where 100% of the load in on the entire line). In this case, unlike in the spliced loop case, the loss of strength at the bend directly transfers to the whole system. So a 1:1 bend will reduce the system by 50% and a 4:1 by 25%. usually 3 or 4:1 is the recommended minimum.

(c) A moving, end-loaded line over a sheave (like a halyard).  The movement creates extra damage to the rope and the bend radius thus needs to be larger. 8:1 is usually recommended.

I conducted the following series of tests on bend ratio and throat ratio:

Test results for 12mm heat set dyneema (NER STS-HSR 75):

Base line test run – to determine strength of this particular batch of line 

(10 samples with splices both ends, with 5:1 bend ratio and 5:1 throat angles)

Average 105% of line rated strength, Standard deviation 5%


Bend ratio run (10 samples at 3 bend ratios, all with 5:1 throat angles)

1:1 Bend Average 94% of base line strength, Standard deviation 4%

2:1 Bend Average 99% of base line strength, Standard deviation 4%

3:1 Bend Average 100% of base line strength, Standard deviation 3%


Throat Angle run (10 samples at 3 throat angles, all with 5:1 bend ratios)

1:1 Throat Average* 83% of base line strength, Standard deviation 5%

2:1 Throat Average 95% of base line strength, Standard deviation 4%

3:1 Throat Average 102% of base line strength, Standard deviation 3%

*Probably a little higher than 1:1 because of splice stretch but as close as I could get it


Used Rope (and sewn splices)

I have tested sewn splices in used/old rope.  The below pic is of some old used Dacron double braid - the kind of stuff that is pretty near impossible to do a normal splice on. Note: Samson has a specific 'old rope' splice, but even it is a bit of a pain to construct in old stiff rope.

1. I tested two sewing patterns.  The top piece has a pattern of three stitches ('straight thru the middle' stitch and then a 'round turn' around each edge stitch) repeated 4 times, for a total of 12 stitched (in doubled v138 thread).  The bottom piece just has 12 'straight thru the middle' stitches.

Top one broke at 560 lbs, and bottom one at 445.

rope splice

I tested the same with 6 sets of stitches in 80lbs test Dyneema fishing line and there was an even greater advantage for the round turn pattern: 1315lbs to 600lbs (for the straight stitch). 

2. I broke a piece of this line with two figure 8 loops in it . . .at 5797lbs. Another piece, with two bowlines, broke at 4880lbs. That is initial indication (confirmed below) that the strength of the line is highly variable along its length. But it suggests an average knotted strength 46% of original line strength, way less than the 70% found in new Dacron double braid. Another test found a very consistent knotted 43% of line strength for used line.

3. Using the stronger 'triple round' sewing, I made a loop with 10 sets of 3 'triple round' stitches, which I figured would be stronger than the line.   It broke at 7062lbs. Another sewn loop piece, this time with a reinforcement piece inserted which tends to test at near 100% of line strength in new line, broke at 6510 lbs. Again suggesting high variability, but also significant strength loss. This is 58% of the new rope strength.

The (lower failure) rope tore apart near the start of the stitching. Notice the stitching is intact.  This is the same failure mode we got with the sewn webbing loops, which surprised me a bit since thick rope is so different than thin webbing.

rope 1 rope2

4. I have tested the core and cover separately.

(a) It is very nice that when they are separated, the core and cover are hollow single braids, so I can do (and test) 'full strength' bury splices.

core/cover eyes

(b) An interesting side note from this testing is that the straight bury splices used to splice the core and cover appear to be much more efficient than the more complex double braid splice used to the splice the line as a while. The spliced spliced Core + Cover separately tends to be about 170% of the spliced whole line, but knotted Core + Cover separately tend to be about exactly equal to the knotted whole line. That suggests that the straight bury splices are stronger than the double braid splice.  I think that the total line is generally about 5% weaker than the sum of its parts because of unequal tension between the core and cover but this varies quite a bit between tests (different lines and mfg's).

The core and cover are the same weight (in the common Dacron double braids), so, as is usually stated, should be splitting the line strength. And they do in new line.  In used line the cover degrades a little bit faster than the core, but a big finding of this testing is that the core is protected less by the cover degrade more than we expected.  As an example, in a used line that as a whole was 67% of original strength, the cover was 67% and the cover 56% (with a 5% testing variation).

(c) The knot tests (vs. splices) implies that, in addition to loosing some overall strength, the line is weaker in bends than new line. This is consistent with the fact that several of the core/cover splices loops broke at the pins (which was 1.5:1 bend radius) where as a new line will never break there.

Sewing Webbing

To summarize all the web sewing details below: There are five practical key messages: (1) many common 'strong looking' stitching patterns come nowhere close to meeting the ORC tether or jack line requirements.  This includes many commercial tethers and jacklines.  Just as one example (others are below) -here is a report on a commercial tether that failed during a Sydney to Hobart race at only 30% of the required minimum load.  (2) There are two ways to make a sewn webbing joint strong enough to meet the ORC requirements. The 'sling' construction (details below) is by far the best structure, but the "internal reinforcing" detail will do the job (but less strong than the sling) if you don't want to make a sling, (3) bar tacking, even 'pseudo-bar tacking' (using short zigzags on a home machine) is the best sewing pattern. (4) The strongest thread possible is best, usually that is V138. Anything less than V92 is unacceptable.

Doing anything other than those later 3 points is extremely likely to generate a webbing loop that will break at well less than the ORC requirement.

1. When sewing two lines or webbing together there is a significant component of clamping friction.  A sewn loop will be somewhere between 115% to 200% of the tensile strength of the thread used.


Hand sewing (breaking at % of stitching tensile strength) 1/4" Dacron double braid 1/8" Dyneema single braid 1" tubular nylon webbing
10 stitches of doubled V69 thread 10.6lb test doubled) 208% 118% 184%
15 stitches of doubled V69 thread 195% 116% 170%
5 stitches of Dyneema finishing line (80lbs test) 150% 140% 125%

2. I found someone else who had machine sewn and broken a lot of webbing samples, but had not done much to analyze his data. The overall summary number is that his samples show an average loop breaking strength = 125% (27.6lbs breaking strength/22lb thread tensile strength) of the thread tensile. However there is very significant variability. The 95% confidence interval is 110% to 140%.

The very highest stitch count samples are all 'under strength', suggesting you will reduce stitch efficiency if you jam too many stitches in. There was no correlation at all with the sewing pattern/shape (e.g. box vs. bar tacks, etc).

sewing strength

I conducted a series of tests to identify the component strength of different types/patterns of stitching.  I tested three variables: type of stitch (straight vs. zigzag), stitch orientation (across the webbing vs. down), and stitch count (20 stitches vs. 40 stitches).  The summary results are:

 Across= 29.0/stitch vs. down = 28.7/stitch - no difference

 Straight=28.3/stitch vs. ZZ=29.4/stitch - a small difference favoring zigzag

 20 stitches=36.8/stitch vs. 40 stitches=32.1/stitch - diminishing marginal returns

The highest results (39lbs/stitch) came from the 20 stitch zigzag samples (the 'across' and 'down' samples were equal), which I shaped to look as much like 'bar tacks' as my sewing machine can make.

3. I draw four conclusions from this testing with regards to making the best webbing loops (as at the ends of jacklines).

(a)  It is easy to make some webbing stitching that "looks strong" but does not meet the 4,500lb minimum requirement for Jacklines.  The sample below is 5" of 9 rows of zig zag, in V69 thread (10.6lbs tensile).  It "looks strong", but breaks at 2,555lbs (63% of the jackline standard).  So, this sewing pattern would meet the jackline requirement if it was 8" long (in this thread), or if done in V138 thread (and still 5" long).


Before After - all stitches broken, on one side Non-broken side
before stitching broken stitch unbroken stitch

and several commercial jackline mfg's are producing stitching that 'look strong' but probably do not meet the ORC requirement. Of the four shown below, only the spinlock is using the proven mountain climbing bar tack.  The other three are using patterns which do not efficiently deal with the load.  Note: The SailRite stitching and webbing I have been able to replicate almost exactly because they publish DIY instructions.  The others I am copying simply from what I see in the picture and it is possible they are using heavier thread (V138 is available, 50% stronger than what I used), so their breaking strengths "could be" 50% stronger but that would still leave two of them (West and Wichard) potentially short of the ORC target.       

SailRite (DIY instructions) West Marine Wichard Spinlock
A replica I sewed broke at 1100lbs A replica I sewed broke at 2140lbs A replica I sewed broke at 2700lbs Webbing broke
sailrite wm wichard spinlock


Again, note this report on a failed commercial tether at exactly the loads my testing suggests.

b. The loads on the stitching on an 'end of webbing loop' are not evenly distributed.  The loads concentrate on the stitching that is at the 'webbing end' (where the loop ends and it steps down from two ply to only one ply).  If you make a long sewing pattern (like say the West Marine above), you might calculate that you have more than enough total stitches, but in fact the stitches toward the loop will carry relatively little load initially, and the stitching will peel up/break from the 1ply/2ply end.

c. So, there are three important considerations when designing a stitching pattern for an "end of webbing loop".  First you want to concentrate all the stitching into as short an area as possible, so that the load up as evenly as possible; second you want to get as effect as possible a set of stitches right at the end as possible; and third you what to use as heavy/strong a thread as possible in order to make each stitch count.  The standard tested mountain climbing solution to this is 5 bar tacks relatively close together.

Bar tacks are best made by specialty bar tacking machines. But you can do a very close approximation with a home zigzag machine.  You want to set the stitch length as short as possible and the zig width as wide as possible.  Below on the right shows one pass of the zigzag, and on the left three passes of the 'pseudo bartack'.  On the left is 16 stitches in about 5/8", or about 25/in.  The width of the zig is about 4mm (note: I should ideally bring the bartack ends closer to the webbing edges)

bar tack  

With 5300lb webbing, I tried the two previous two strongest patterns (double Box X, and Bar Tack) with V138 thread.  I should note that my machine does not like this thread and webbing combination.  The back side of the stitching was sloppy looking and I am going to fiddle around with the settings to see if I can make it better . . . But I got my first stitch pattern that passes the ORC specification (4500lbs) - the standard mountain climber 6 bar tacks (note: these were not 'proper' bar tacks, but short zigzags).  The stitching did not break, the failure was the webbing broke right in front of the first bar tack.  The Double box X broke at 3710.  And a water knot broke at 3070 in this webbing. I tried both these patterns in V92 thread (30% smaller/weaker than the V138) and got much nicer stitching but less strong (2460lbs for Double Box X, and 4320lbs for the 6 bar tacks - where the stitching did not fail, rather the webbing failed just at the first bar tack) . .  .the neater stitching did not make up for the 30% decreased thread tensile strength. So far, 6 bar tacks in V138 thread looks like the ticket for OCR compliant jacklines.

bar tack

(d) If your sewing machine can not do zigzag, then packing a lot of short stitching into the last inch (closest to the 1 ply/2ply transition) is a good choice

one inch

(e) It is possible to make really strong individual hand stitches (using doubled 80lb Dyneema fishing line for instance), but to get 4,500lbs on a pattern you need to have enough stitches (about a 100) to spread it smoothly out over heavy polyester webbing.

40 stitches in doubled 80lb test fishing line Stitches pulled right thru webbing
Suggests you need a minimum of 100 stitches to spread 4,500lbs
hand stitching broken hand

5. "End to end" slings are much easier to make high strength than the above "end of webbing" loops.  This is because on an end to end joined sling the join is equally loaded from both sides and the stitches are all (relatively) evenly loaded., whereas on an "end of webbing" loop, the load is all on one ply and the stitches are highly loaded at one end (near the 1 ply/2 ply transition) and low loaded at the other end.  Looking at commercial/industrial slings rated over 3 or 4 tons, they are all "end to end joined slings", and if the application calls for two loops (rather than one sling), then the two webbing parts are just sewn together down the middle of the sling leaving loops on the ends. This construction could be done on jacklines - get some webbing twice as long, sew the ends together using pattern 1 in the chart below, then sew the two 'plys' of webbing together except for the end loops (while keeping equal tension on the two plys - this could be done by sticking them together with basting tape before sewing).  The high load commercial sling products mostly use the #1 stitching pattern below, but the high load mountain climbing slings use bar tacks (pattern #3, except with 5 or 6 bars).  I am not sure why those two similar high load applications use different patterns, expect perhaps they are using different sling material/widths.  Note: more details on the below testing.


end join

Related to this are two studies done on parachute webbing joint construction: USA Parachute, UK Parachute. These studies did not test bar tacks.  But they do offer many results on improving webbing joint strength, including the conclusion in both that adding webbing reinforcing locally to the joint - either strengthening the ends where the loads are highest by wrapping webbing/fabric around the ends of the joint, or spreading the loads at the ends by running webbing longitudinally thru the joint (or over) and past its ends by an inch or two. 

webbing joints


6. So, if you want to make an ORC spec jackline is 1" Dacron webbing, there is a pretty simple sure way to achieve it. The below broke at 4990 lbs.  The only downside is it takes twice as much webbing, but you can use basic stuff and not the special high strength webbing.


(1) Take some pretty basic webbing (this is 2700lb tests stuff and not the fancy 5300lbs stuff)
and sew the two ends together with 5 bar tacks (three passes of tight zigzag each)
(2) Then sew one row of long zigzag down the centerline,
 leaving just the end loops you want 
(on a full length jackline you would want to stick the webbing together
 with basting tape before sewing to get it all even).
sling 1  sling 2 


I tried a version with only 3 bar tacks and it broke at 3060lbs and a version with a double box X and it broke at 4290lbs, so 5 or 6 bar tacks (short and wide zigzags) is the best ticket. The whole challenge of getting 4,500lb stitching does suggest considering using Dyneema jacklines, where 4,500lbs and 100% splice are dead easy to achieve. And it will be less sensitive to chafe and UV (the stitching on the webbing jackline is vulnerable to both).


7. I tested adding the 'reinforcing pieces' suggested in the parachute testing.  I tested both the 'inside reinforcing' approach, and an 'outside reinforcing' approach (the white webbing in the photos below). These reinforcing pieces help by spreading the load at the stress riser point.   In samples above I had great difficultly getting straight sewn loops anywhere near the ORC 4500lbs. The sewing pattern I use here were not the strongest, and probably would have broken at 2500-3500lbs in a 'plain loop'.  But with the parachute reinforcement one of (the 'outside reinforcement') these loops broke at 4570lbs, and the other ('inside reinforcement') was stronger. So, while this technique is not as strong as the 'sling' technique just above, it is much stronger than plain loops and does get you to the ORC minimum, with less sewing effort and less webbing length (although you do need the extra strong webbing for the base here, unlike with the sling approach). This was only one sample, but the "inside" approach was both easier to make and stronger, so I think I would stick with that.


web 1  web 2 

Out of curiosity I did a bake-off between the commercial West Marine (the actual branded product) jackline loop vs. the above 'parachute reinforced' loop.

west1 west2
west3 west4
west5 It is pretty clear that either the parachute approach (with excellent webbing) or
 the sling approach (with more normal webbing) are the way to go for making these
 loops. The west sewing patterns is commonly seen on industrial slings, but in a sling
 application the threads are more evenly loaded that they are in an 'end loop'.

Putting all the above lessons into practice, I made a set of jacklines. It is a 'sling' design which was by far the most efficient/strongest in testing.  This means it is a double length of webbing, with the ends sewed together in the middle of the sling.    The ends are joined in the middle of the jackline.  This area is thus 3 layers of webbing sewed with 7 bar tacks.  If I were doing this for a cruiser, who would leave the jacklines set for long periods, I would cover this stitching with a light taffeta to protect the stitching. At the end loops I added extra webbing for 'wear' reinforcing, because the loops get the most wear when lashed or cow hitched.  And between the middle join and the loops, it is a double layer sewed with continuous zig zag.

middle join between middle and loop
end loop loop detail

7. Knots in webbing - there are two commonly used in flat (non-tubular) webbing, the water (or overhand) and the figure 8. The figure 8 is a 10% stronger (in flat polyester webbing) and not much more difficult to tie


Water (overhand) Knot Figure 8 Knot
1540lbs breaking (57% of webbing tensile) 1910lbs breaking (71% of Webbing tensile)
water knot figure 8


8.Sails use a lot of adhesive today for construction, so I thought it worth testing whether adhesive could be useful for webbing.  Net net, the answer is no. Use water knots for fast and easy loops, and (lots of) sewing for full strength loops.

Test results (loops bonded with 5 sq in of adhesive)

-3M 5200 Fast cure: 191 psi, Stdev 5

-3M Spray 90 (High Strength): 110 psi, stdev 11

-3M Spray 80: 78 psi, stdev 13

For reference:

-Sewing 9 rows of zigzag: 567 psi, stdev 18

-Water knot: 511 psi, stdev 4

9. Quickly sewn 'low load' webbing loops are often used for attaching sail slugs and potentially for cat tramps.  I tested the simpliest approach - two loops of inexpensive polyester webbing (2700lbs loose weave stuff), with four stitches of 80lb fishing line thru the middle.  The stitches pulled thru the webbing at 1370 lbs.  This could obviously be made much stronger with either more stitches or 'firmer weave' webbing. But 1370lbs is probably enough for most of these applications.



I tested a range of thimble designs.  The loads quoted are all when a wall of the thimble started to distort sideways. They have already collapsed and elongated to some degree by this point.

There were two surprises.  First is how poor the conventional open galvanized thimble (as used on many anchor rodes) is.  Both the 5/16" and 1/2" sizes distorted at only 57% of the tensile of the appropriate size nylon 3 strand line. Second is how relatively good the closed nylon thimbles were (compared to the galvanized).  They distorted at slightly higher load, and bounced back to near their original shape when the load was taken off. The open stainless thimble distorted at essentially 100% of the line tensile, so is "good enough".  The clear winner is the closed stainless thimble (aka a sailmakers thimble).  It was the only one which exceeded the nylon tensile strength - giving a safety margin for nylon and allowing for use with higher strength lines.

All tested thimbles, after testing Plastic thimble before loading Plastic thimble under load
thimbles plastic1 plastic2


Most grommet applications are not 'high load', but they are sometimes used on multi-hull trampoline edges.  So the question was what sort of load can they hold and how strong does the lashing need to be.  I tested stainless spur gromments in 4 layers of 18 oz Shelter-rite (a common tramp border material).  The pull out load was 340lbs.

stainless grommets


1. In a test of rescue block efficiency, it was determines (not too surprisingly) that bearing type and sheave size both affect pull friction.  In the graph below, note the 7% increase in efficiency between the similar CMI blocks as you move from 1.5" dia to 3.75"dia. And note the 6% increase with the similar CMI blocks as they moved from a bushing to needle bearings. Finally, when thinking about 'low friction rings' look at how poor the .5" carabineer is. Here is another study on Arborist pulley efficiency. And here is another good study of block friction.

block efficiency

Below is the results of some similar testing I did to look particularly at "low friction rings" and carabineers.  The results are not surprising - if friction is important, get the biggest sheave and ball bearings.  I tested different brands/models of similar size ball bearing blocks and could not measure a statistically significant difference.

block friction 

2. Smaller line helps efficiency (note the gain in the static-pro line as you move from 1/2" to 3/8"). And somewhat surprisingly to me, wire is quite efficient.  Here is another study on Arborist rope efficiency.

wire vs rope

3. Friction on a 'low friction ring' will depend on the deflection angle thru the ring. It is mostly dependent on the 'load factor' (180 degree deflection = 2x load factor, 0 degrees = 1x load factor) on the block (the 'expected' line in the graph below). But there is an additional smaller factor related to internal line friction.


4. When you look at line size vs. low friction ring bend radius you have a steady increase in friction as you decrease bend radius (e.g. make it a tighter bend), until you get below a 1:1 bends (where the line is bigger in dia than the ring).  Then the friction goes up exponentially (data).

All test pulls: friction vs. bend radius (hardware dia/line dia) Average of 5 pulls for each line size/hardware combination.
Average of a 44% increase in friction going from the antal ring to the carabineer.
bendfriction5  frictionbend2

Dyneema Life lines

I have an extensive best practice article on Dyneema life lines.

The following is a ball park calculation of the loads on life lines with people 'fully hiking' on them. You obviously don't want to exceed 100% of the line breaking strength, but some would say you want a safety factor and should not exceed 50%.

hiking loads

Related to life lines I have done some chafe testing, and one finding is that chafe in dyneema is much worse than it looks - eg the line loses more strength than one would think looking at the line.

This line was 23% weaker than rated This line was 64% weaker than rated
chafe1 chafe2a


 I have done some testing to explain why this is.   I took a piece of amsteel (a line made of 12 strands) and cut half the fibers in each of three of the strands. That cuts 12.5% of the fibers, and damages 25% of the strands.   I then broke it. The line broke at 23% below tensile.   So, it appears to be the fraction of strands damaged that is important rather than the fraction of fibers. Explanation: The damaged strands cannot hold their rated proportion of the load and thus fail and then the rest of the line fails.

Wire seizings

This testing is all with .041 aircraft stainless lockwire.

The 'conventional method' - two loops and then twist
 the ends 8 times - the twist comes apart at 450lbs

In order to reduce the load on the twist, I then tried taking two
turns around the seizing before twisting the ends.  But this does
 not allow the seizing legs to equalize load and the wire broke at 420lbs.
In order to allow more equalization I just took one turn around
the seizing before the twist.  This was better - wire broke at 470lbs
s2 s3

In order to allow complete equalization while still taking some
 load off the twist I took a turn around one end before the twist.
 This broke at 500lbs - it is the best structural solution.
Then I did the same but with 3 loops - broke at 840lbs
s4 s6



Not completed yet:

Bollard pull

Anchor and mooring loads

Rig loads

Data Sheets

Price & specs summary/comparison (excel spreadsheet - published specs and retail prices)

Samson Amsteel Blue (Dyneema single braid)

Samson LS (polyester double braid)

Samson Pro-Set (nylon 3 strand)

Samson Superstrong (nylon double braid)

Yale Nylon Brait (nylon braid - 8 strand)

New England Ropes Stay-Set (polyester double braid)

New England Ropes STS12 (Dyneema single braid)

New England Ropes WR2 (Dyneema double braid)

Related Stuff

Instructions for the knots I use, but infrequently enough to want some instructions

Inspection criteria for HMPE rope

Abrasion and twist effects

Effect of heat on Dyneema strength

Knot strength

Breaking strength of various fishing knots

Dyneema abrasion protection

Dynamic Fatigue

Dyneema Life lines

Dyneema Splice test videos

Hurricane mooring systems

Jacklines and Tethers

Mooring with HMPE

Sewn Eye Splices

Tensile behavior of mixed fiber line

Tests of high strength climbers rope

Testing of high strength rope

Effect of bending on tensile strength

Water knot slipping

When should a tow line be retired

Kermantle eye splice

ASTM D4268 standard

Rope Material Properties

Ashley Book of Knots in pdf (very big)

Bury splice instructions

Single Brummel Splice instructions








"What is freedom? We say of a boat skimming the water with light feet, 'How free she runs,' when we mean how perfectly she is adjusted to the force of the wind, how perfectly she obeys the great breath out of the heavens that fill her sails. Throw her head into the wind and she how she will halt and stagger. She is free only when you have let her fall off again and have recovered once more her nice adjustment to the natural forces she must obey and cannot defy."

Woodrow Wilson