Because the force is large and the cable is polymer, so concentrating the grip would crush it.

A dead-end holds the full tension of the span - potentially several kilonewtons under ice loading. On a steel messenger that force can be taken over a short length because steel tolerates the pressure. On ADSS or on the jacket of a fibre cable it cannot: the same force applied over a short length would deform the jacket, damage the buffer tubes and eventually let the cable pull out.

Distributing the grip along a longer length reduces the pressure at every point to something the cable tolerates while the total holding force stays the same.

That is what preformed helical dead-ends do. A set of shaped rods is wrapped around the cable over a substantial length, and the grip develops progressively along that length - the more the cable is pulled, the tighter the rods wrap.

It is also why a dead-end is cable-specific: the rod set is designed for a particular diameter and construction.

By wrapping shaped rods around the cable so that tension tightens the wrap - a grip that increases with the load rather than being fixed at installation.

The dead-end is supplied as a set of pre-curved rods joined at a loop or thimble at one end. The installer wraps them onto the cable one at a time or in sub-sets, following the pre-formed helix.

Once wrapped, pulling on the loop tries to unwind the helix against the cable, which makes the rods grip harder. The harder the cable pulls, the tighter the grip - a self-energising action, and the reason a correctly applied preformed dead-end does not slip.

The grip is spread over the whole wrapped length, so pressure at any point is low.

The critical requirements are that the correct rod set is used for the cable diameter, that the full length is wrapped, and that the rods are wrapped in the correct direction and fully seated. A partially applied dead-end grips at first and slips later.

At terminations, at sharp angles, at the ends of tensioned sections, and wherever the cable must not move.

The obvious case is a termination: the cable ends at a pole and goes down to a joint, a building or a cabinet. There is no span on the other side, so the whole tension has to be held.

The less obvious case is an angle. A suspension clamp copes with a small change of direction, because the horizontal component is modest. Beyond a defined angle the resultant force is too large and the cable will pull through, so the correct arrangement is a dead-end on each side with a short jumper between.

Long routes are also divided into tensioned sections by dead-ends at intervals, so that a failure or a repair in one section does not release tension along the whole route - and so that sag can be controlled section by section.

Steep changes in level and crossings of roads and railways are usually dead-ended for the same reason.

The span's design tension under worst-case loading - which is not the tension at installation, and is usually a good deal higher.

The cable is tensioned when it is strung, at whatever the ambient temperature happens to be. In service the tension rises when the cable contracts in cold weather, and rises much further when the span is loaded with radial ice and pushed by wind.

The governing case is therefore the combination defined by the local loading standard, and it can be several times the installed tension.

The dead-end must hold that with a safety factor, and so must everything behind it: the bracket, the pole band and the pole itself. A dead-end that is adequate attached to a bracket that is not simply moves the failure.

So size the whole load path from the sag-tension calculation for the span, not from what the cable feels like when it is being installed.

With hardware designed for it, because the tension is carried entirely in the cable's own jacket and aramid members - there is no messenger to grip.

On a messengered cable the dead-end grips the steel messenger, and the cable itself is lashed to it and carries no tension. That is mechanically straightforward.

ADSS has no messenger. The strength members are aramid yarns inside the cable, and the only surface a clamp can grip is the outer jacket. So the dead-end has to transfer the entire span tension into a polymer surface without damaging what is underneath.

Preformed helical dead-ends made specifically for the cable's diameter and jacket are the standard answer, wrapping over a long length so the load is distributed.

On shared power routes there is a further requirement: the fitting must manage the electrical field at the cable surface. In high field regions, dry-band arcing can erode an ADSS jacket, and dead-end hardware for those positions is designed to control it.

Wrong rod set for the cable, incomplete wrapping, reuse, and wrapping in the wrong direction - all of which grip initially and fail later.

Using a rod set for a different diameter gives either too little grip, so the cable creeps out over months, or too much local pressure on a smaller cable.

Incomplete wrapping is the most common: the installer stops short of the full length, or does not seat the ends properly. The dead-end holds at installation tension and slips when ice loading arrives, which may be a year later.

Reuse is a real temptation on a repair and should be refused. A preformed dead-end takes a set when it is wrapped; unwrapping and rewrapping it produces a grip that cannot be relied on.

Wrapping against the cable's own lay direction reduces the grip substantially.

All four failures share the same signature - the route was fine when it was built and a span went slack later - which is why installation practice matters more here than component choice.

Material and coating matched to the environment, and attention to the thimble, bracket and bolt as much as the dead-end itself.

Preformed rods are usually galvanised steel or aluminium-clad, chosen for the atmosphere. Coastal and industrial environments call for stainless or a heavier coating, and the choice should follow the local exposure rather than a general specification.

The items that actually fail are usually the attachments. The thimble or loop at the end of the dead-end, the shackle or bolt connecting it to the bracket, and the pole band are all steel components carrying full tension in the open air, and a corroded bolt drops the cable just as effectively as a failed clamp.

Dissimilar metals are worth watching: a stainless fitting against a galvanised bracket in a wet coastal atmosphere corrodes the galvanising preferentially.

Specify the whole load path in compatible materials, and include the attachment hardware in the inspection rather than looking only at the dead-end.