Two drums of the same reference colour are rarely identical. On a large facade the difference reads as a repair, not as a colour, and no amount of extra paint fixes it.
Colour complaints are the most expensive kind of complaint to resolve, because the remedy is rarely a small one. A slightly different drum of the same reference is acceptable on a hidden steel beam and unacceptable on the visible face of a building, and the difference between those two outcomes is decided long before anyone opens the drum.
The reference is a target, not a formula for identical drums
A colour reference such as RAL, BS or NCS defines a target within a tolerance. It does not define one exact point, and it does not commit two production batches to the same point within that tolerance. Pigment lots differ, tinting pastes age, and the same reference produced in two plants can sit on opposite sides of the allowed window. Both drums are correct, and both are different.
This is why ordering the full quantity for a structure in a single batch, with a single batch number recorded on the delivery documents, is standard practice on visible work. Where the quantity exceeds one batch, the specification should require sequential batch numbers and a colour check between them before application begins, not after the last coat.
What a difference means numerically
Colour difference is expressed as Delta E in a defined colour space. As a rough guide, a trained eye begins to notice a difference in the region of one unit, an untrained observer somewhere between two and three, and beyond three the difference is obvious on a large flat area in daylight. These figures are indicative; the perception depends on gloss, surface texture and lighting, and on whether the two surfaces meet at an edge.
Sheen is the variable most often ignored. Two coats matched in colour but differing in gloss will read as different colours on a wall, because the gloss changes the proportion of light reflected specularly towards the viewer. It is common to find a repair that is a perfect colour match and still visible, purely because the touch-up was applied thinner and dried duller.
Metamerism makes it worse
Two samples can match under one light source and diverge under another. This is metamerism, and it appears when the pigment composition differs even though the measured colour under the reference illuminant agrees. A patch checked under sodium lamps at night and approved can look wrong in morning daylight. Visual assessment should be made under a defined illuminant, and on site that usually means north daylight or a controlled lamp, not whatever is available.
The repair problem
Touching up a small area with a different batch, or with the same batch applied by a different method, almost always shows. The eye is extremely good at detecting edges, and a rectangle of slightly different colour on a large flat surface is an edge. Where a repair is unavoidable, the practical rule is to break the repair at a natural boundary: a corner, a joint, a change of plane, or the full width between two features. Coating a whole panel costs more material and less credibility than a patch that reads as a patch.
What to put in the specification
- Single batch for visible areas, with batch numbers recorded and retained
- Delta E limit agreed in writing, with the measurement geometry and illuminant named
- Sample panels retained from each batch, stored indoors, signed and dated
- Gloss range specified alongside colour, not separately
- Repairs broken at natural boundaries; full panel where no boundary exists
- Visual acceptance in daylight, not under site floodlights
The cheapest control is the least popular one: buy the whole visible area in one order, and keep a panel from it. That panel will settle more arguments than any instrument reading, because it is the only sample that has aged in the same place as the structure.