There were four different cornice profiles on one building, and the owner wanted to know which was the original. Three of them were, in a sense. The building had been repaired in 1931, again after a fire in the sixties, and once more by somebody with a router and an afternoon. Before anybody replicates a piece of an old facade, the building has to be read: which material is original, which is a repair, what the repair was trying to solve, and whether the profile in front of you is a copy of a copy. That reading is a survey job, and it happens before a single template is cut.

Date the materials, not the building

A facade is a stack of decisions made across a century, and each one leaves a material signature. Lime-based mortar with a soft, buff-coloured matrix and visible lumps of unburnt lime generally predates the widespread use of Portland cement; a grey, dense, hard mortar is later. Terracotta detail was pressed and glazed, and its back face is hollow and ribbed; cast stone of the same period is solid, with a visible aggregate when chipped. Sheet-metal cornices — galvanised steel or copper worked over a wooden former — are common on turn-of-the-century commercial fronts and are often mistaken for stone from the sidewalk. Knowing which one you are looking at decides everything downstream, because a terracotta unit cannot be patched with mortar and a sheet-metal cornice cannot be pinned like stone.

Survey before specification

A close-up survey means hands on the material, not binoculars. In New York, buildings over six storeys fall under the cyclical facade inspection programme that grew out of Local Law 11, which means there is usually a filed report describing conditions, and that report is the first document to read — it will tell you what a qualified examiner saw at the last cycle and whether any condition was classified as unsafe. For the detail work itself, a drop or a lift gives you what matters: sounding the material with a hammer to find delamination, checking whether cracks follow the joints or run through the units, and looking for rust jacking where an embedded steel anchor has expanded and split the stone. Where the report and the survey disagree, facade repair specialists in New York and a preservation-minded joinery or stone shop should be in the same conversation before scaffolding comes down, because the repair sequence changes what can be replicated.

Deciding what to replicate and what to consolidate

Not every damaged element should be replaced. A weathered profile that has lost its crisp arris but is structurally sound is usually better consolidated and left, because the replacement will read as new for twenty years. The honest test is whether the unit is still doing its job: shedding water, carrying load, holding its anchorage. Where a unit has failed — a cracked-through sill, a spalled bracket with exposed rebar, a rotted timber modillion — replication is the right answer, and it should be replication in a material compatible with its neighbours. Putting a dense Portland-based cast piece next to soft original stone concentrates stress in the old material and quietly accelerates its decay.

Templating a profile that no longer exists whole

The method has not changed much. Find the least-weathered surviving example, take a profile gauge reading at several points along it, and back that up with a plaster or silicone squeeze of a short length. Photograph with a scale rule in frame from square on, then correct for lens distortion. Where every example is worn, the profile can often be reconstructed from the logic of the period: classical mouldings are combinations of a small number of curves, and the missing radius is usually recoverable from the parts that survive. Historical photographs and insurance atlases are worth the archive time, and so is looking at the neighbouring buildings, which were frequently built by the same hands from the same catalogue.

What the work costs and how long it takes

Survey and documentation on a modest storefront or a single bay usually runs a few thousand dollars before any repair is priced, and that money buys the drawings that keep the repair from being invented on site. Replicating a carved or cast detail is priced per unit and depends on whether a mould can be reused: one-off carving is measured in bench days, while a run of eight identical brackets from one mould drops the unit cost sharply. Lead times of eight to sixteen weeks for replicated masonry units are normal, which is why the survey has to happen before the scaffold licence, not after.

Frequently asked questions

Can a missing detail be reproduced from photographs alone? Sometimes, if the photographs are square-on and contain a known dimension for scale. A physical profile taken from a surviving element is far more reliable and should be sought first.

Is it acceptable to replicate stone detail in cast material? Frequently yes, and it has been done since the nineteenth century. The requirement is that the substitute is compatible in strength and vapour behaviour with the surrounding masonry, not merely similar in colour.

Should paint be removed from an old facade before survey? A test patch, yes, because paint hides both the material and its condition. Wholesale stripping before anyone knows what lies underneath risks damaging soft stone or glazed terracotta.