The shape of the embedded end - a J curves back on itself, an L turns a right angle - and the J generally develops slightly more resistance for the same bar.

Both work the same way: the bend bears against concrete and resists the bar being pulled straight out.

The J's return hook engages more material and is harder to straighten, which is why it is the more common of the two in tension applications.

The L is simpler to fabricate and takes less space in a congested reinforcement cage, which matters in thin sections.

Neither is as predictable as a headed anchor with a plate or forged head, and design codes treat hooked bars conservatively for that reason.

Where the design specifies one shape, substituting the other changes the assumed capacity.

Enough for the grout bed, base plate, washer, nut and thread run-out - and it is worth calculating rather than estimating.

The grout or levelling layer under a base plate is often thicker than expected and is the component most frequently forgotten.

The nut must engage fully, which means the thread should reach at least the top of the nut and preferably a little beyond so engagement is visible.

Levelling nuts under the plate, where used, add their own height and must be included in the total.

Excess projection can be cut, but cutting damages the thread and needs a die to recover it.

Insufficient projection has no easy remedy at all, which is why erring long is standard practice.

By a rigid template that fixes both spacing and height, tied so it cannot move under vibration.

A template plate drilled to the final hole pattern is the reliable method. It guarantees the group matches the base plate and proves the pattern before concreting.

The template is fixed to the formwork or to a frame, not simply rested on the reinforcement.

Poker vibrators move bolts that feel securely tied, so restraint has to resist vibration rather than just weight.

Positions should be checked immediately after placing, while the concrete is still workable and a small correction is possible.

Recording the as-built positions before the steel arrives lets any discrepancy be resolved off site instead of at erection.

Slot or enlarge the base plate if the error is small; otherwise abandon it and design a post-installed replacement.

Slotting is the normal and cheapest remedy, and it is why plates are frequently detailed with slots from the outset.

Bending a bolt into position is common site practice and a poor one - it yields the steel, changes the load path and cannot easily be justified afterwards.

A post-installed anchor nearby needs its own check for edge distance and for spacing from the abandoned bolt.

Where a whole group is out, a larger base plate or a bolted-on adaptor plate is sometimes the practical answer.

Any of these is a decision for the engineer responsible for the connection, not a site remedy.

Less so than headed anchors - a hook can straighten and pull out, and codes treat hooked bars conservatively.

The hook resists withdrawal by bearing on concrete and by its own resistance to being straightened, and both are less predictable than the bearing of a head or plate.

In high tension the failure can be progressive: the hook opens slightly, capacity falls, and the bar continues to draw.

Headed studs, bolts with a nut and plate washer at the embedded end, and cast-in assemblies with anchor plates are the alternatives where tension governs.

Concrete around the hook must be well compacted, since a void behind the bend removes the bearing entirely.

Some design codes restrict or discount hooked anchors in seismic applications for these reasons.

With tape, a plastic cap or a sacrificial nut - and it takes far less time than cleaning a fouled thread later.

Concrete splashes onto projecting threads and sets there. Once hard, it has to be removed mechanically without damaging the thread beneath.

A sacrificial nut run down the thread before the pour protects it and can be run off afterwards, clearing any residue as it goes.

Plastic caps do the same and are quicker, though they are easily knocked off in a busy pour.

Tape is the least reliable but is better than nothing.

A thread chaser or die run over each bolt before the steel arrives is cheap insurance even where protection was used.

Enough for the exposed end and the surface interface, because the embedded length is already reasonably protected.

Sound alkaline concrete passivates embedded steel. The problem is where the bolt breaks the surface and where water can track down beside it.

Hot-dip galvanising is the general external answer, and galvanised bolts need oversize-tapped nuts to clear the coating.

Where exposure is severe, or where nobody will ever reach the bolt again, stainless is the answer - and near the sea it needs to be a molybdenum-bearing grade.

Grouting the base plate and detailing to shed water off the fixing does more than a material upgrade in many cases.

The nut, washers and plate should be compatible with the bolt to avoid setting up a galvanic couple at the most exposed point.