Every coupling step on solid phase leaves behind some fraction of chains that did not react. You cannot see them, you cannot weigh them, and if you carry on regardless they become deletion peptides or, if you cap, truncated ones. The colorimetric on-resin tests exist to answer one question before the next Fmoc removal: is there still a free amine on these beads? What they actually report is narrower than that question, and the gap between the two is where most of the trouble sits.
What the ninhydrin reaction is telling you
Kaiser, Colescott, Bossinger and Cook published the test that carries Kaiser's name in 1970, as a colour test for the detection of free terminal amino groups in solid-phase peptide synthesis [1]. In practice it is three solutions added to a few washed beads and heated: ninhydrin in ethanol, phenol in ethanol, and potassium cyanide in pyridine.
The chromophore is Ruhemann's purple, the dye formed when ninhydrin reacts with a primary amino group, discovered by Siegfried Ruhemann in 1910 [2]. Friedman's survey of the reaction states the specificity problem exactly: at pH 5.5 the reaction produces the same soluble chromophore from all primary amines that react, whether amines, amino acids, peptides, proteins or ammonia [2]. The colour does not identify the amine. It reports only that a reacting primary amine was somewhere in the tube.
The same review notes that the chromophore is not chemically bound to the insoluble material and so is not lost when that material is removed by filtration or centrifugation after the reaction [2]. Sarin, Kent, Tam and Merrifield took that further while building the quantitative form of the assay: under their controlled conditions all of the blue chromophore is found in solution and the polystyrene beads themselves are colourless, and they proposed an ion-exchange mechanism to account for the blue beads generated by the earlier qualitative procedures [3]. On that account the blue bead you are squinting at is adsorbed dye — one inference removed from the quantity you actually want, which is the number of unacylated chain ends.
Proline reads clean whether or not it is
A secondary amine is different chemistry with a different product. Friedman lists imines such as proline and pipecolic acid among the compounds that react with ninhydrin to give chromophores other than Ruhemann's purple [2]; that survey concerns free compounds in solution, but the consequence on resin is the one that bites. A peptidyl-resin whose N-terminal residue is a secondary amine does not develop the diagnostic blue, so an observer scoring blue against not-blue extracts no information from it. A proline coupling returns the same visual verdict whether it went to completion or stalled at half. The same holds for any N-alkylated residue.
Kaiser's group published a separate colour test for terminal prolyl residues in 1980 [4]. A year earlier, Christensen and co-workers had introduced chloranil (tetrachloro-1,4-benzoquinone) as a qualitative test for monitoring coupling completeness on solid phase [5]. Vojkovsky re-examined the chloranil detection of resin-bound secondary amino groups in 1995 and reported a procedure roughly ten times more sensitive than the original: every secondary amino group examined except N-(p-tolyl)glycine was detected reliably down to a substitution of 2.8 microequivalents per gram, and protected imidazole in His(Trt) did not interfere [6]. A later reagent covers both classes in one operation — Claerhout, Ermolat'ev and Van der Eycken's DESC salt generates a coloured azadiene on the bead and detects resin-bound primary and secondary amines, with protocols that allow primary amines to be read in the presence of secondary ones [7].
None of this is optional bookkeeping. If your sequence contains proline or an N-methylated residue and your monitoring is the ninhydrin test alone, those are exactly the steps whose completion your monitor cannot report on at all.
Where the floor sits
The qualitative test resolves whatever your eye resolves. Sarin and colleagues addressed that by moving the readout off the bead entirely: the free amine is reacted with ninhydrin under carefully controlled conditions and the resulting chromophore is determined in solution at 570 nm, with a precision of plus or minus 4% across the range 0.005 to 1 micromole, which makes it possible to establish when peptide bond formation is more than 99.9% complete [3]. That is the sensitivity regime a long synthesis actually needs, and it is reached by spectrophotometry, not by inspection.
The same assay supplies the denominator. Sarin and colleagues describe it as useful for measuring the total number of peptide chains on the resin and, especially, for monitoring the progress of the coupling reaction [3]. That distinction matters more than it looks: a residual-amine figure without a measured loading gives you an absolute amount of free amine but not the completion percentage you are actually trying to control.
Tests that quote a floor quote it as an absolute loading. Hancock and Battersby's micro-test with 2,4,6-trinitrobenzenesulphonic acid detects 3 nmol of free amino groups per milligram of resin [8]; the improved chloranil procedure reaches 2.8 microequivalents per gram [6]. Both sit near 3 micromoles per gram. Set that against a resin loaded at 0.5 mmol per gram and the floor is around 0.6% of sites — which is roughly the failure rate that destroys a long chain. Fifty couplings at 99% each leave 60.5% of chains full-length; at 99.5% each, 77.8%; at 99.9%, 95.1%. A negative from a test whose floor sits near 0.6% of sites is compatible with any coupling above roughly 99.4% complete, and that band still contains the difference between a fifty-mer that survives and one that does not.
Gaggini and colleagues state the general position in their survey of colour tests for solid-phase organic synthesis: such tests, far from providing the ultimate solution, may help in detection and sometimes in quantification of functional groups [9]. Read the qualitative Kaiser test as a coarse alarm — worth acting on when it screams, provided the beads were washed first, and uninformative when it whispers.
What produces a wrong answer
Carried-over reagent. Because every reacting primary amine yields the same chromophore, ammonia included [2], anything bearing a free primary amine still wetting the beads reports as though it were an unreacted chain end. That is why the sample must be withdrawn and washed properly before the reagents go in. Piperidine and the tertiary bases used in activation sit outside the primary-amine chemistry the colour depends on; free amino acid and ammonia from any source do not [2]. A positive test on an inadequately washed sample tells you about your washing.
Side chains and their protection. In solution, ninhydrin also reacts with the guanidino group of arginine, the amide groups of asparagine, the indole ring of tryptophan and the sulfhydryl of cysteine, each giving its own chromophore [2]. On resin these groups are normally masked, so the operative question for any given colour test is whether a particular side-chain protecting group stays silent and intact under the test conditions — a question answered per group, not in general. Vojkovsky answered one instance, showing that trityl-protected histidine imidazole did not interfere with his chloranil procedure [6].
The readout is bead colour. The qualitative test asks you to score the colour of a solid, so anything else that colours the beads — a chromophoric linker, a dyed or tagged support, a coloured by-product held in the matrix — sits on top of the signal you are trying to read. This is a structural limitation of the format rather than a defect of the reagent, and it is the reason the quantitative variant reads a solution instead [3].
Sampling is destructive and small. The beads you test are consumed. You are also inferring the state of the whole reactor from a few beads pulled from it, which is fine when the bed is well mixed and much less fine when it is not. Neither cost matters much once per cycle in a manual batch synthesis; both become prohibitive when the cycle time is minutes.
Monitors that need no sample
Krchnak and Vagner monitored completeness of acylation of resin-bound free amino groups with the acid-base indicator bromophenol blue while running fifty peptides on polystyrene in a series of concatenated low-pressure continuous-flow reactors [10]. Colour monitoring in a flow reactor is read during the synthesis rather than on withdrawn beads - an inference from the format rather than a statement in the paper - and it is still a colour judgement.
McFerran, Walker, McGurk and Scott took a different route: adding a sterically hindered base to the aprotic polar synthesis solvent ionises the acids formed in coupling and deprotection, producing a conductance signal in the solvent whose changes let reactions at the resin be followed with no degradation of coupling efficiency [11].
The measurement that instrumented synthesis uses in place of bead tests is Fmoc release. Suspending an Fmoc-substituted resin in 20% piperidine in DMF generates the dibenzofulvene-piperidine adduct, quantified by UV-visible spectroscopy at its absorption maximum of 301.0 nm; molar absorption coefficients reported for that adduct at 301.0 nm range from 7100 to 8100 L per mole per cm, and Eissler and colleagues note that the nominal batch size may differ by up to 14% depending on which value is applied. They redetermined the coefficients at 301.0 and 289.8 nm and point out that measuring at 289.8 nm reduces the impact of wavelength accuracy [12]. That work concerns substitution determination on starting resins rather than per-cycle monitoring, but it is the same deprotection chemistry an in-line detector integrates every cycle, and it fixes the size of the systematic error you inherit from your choice of coefficient.
Flow synthesis leans on it because a cycle time of minutes leaves no room to withdraw and score a bead sample. Simon and colleagues passed the reagent stream through a UV detector for continuous monitoring while incorporating a residue every 1.8 min under automatic control [13]. Mohapatra and co-workers describe the detection directly: Fmoc deprotection is generally quantitative and the dibenzofulvene by-product is detected with an in-line UV-visible detector at 310 nm, and the integral, height and width of the time-resolved deprotection traces can be used to identify mass-transfer problems during deprotection, which they interpret as aggregation on resin. They applied deep learning to 35,427 individual Fmoc deprotection reactions and predicted the outcome of deprotection reactions with less than 6% error [14].
The two readouts are not interchangeable. Eissler and colleagues quantify the dibenzofulvene-piperidine adduct at its 301.0 nm maximum [12]; the flow work describes detecting the dibenzofulvene by-product itself with an in-line detector at 310 nm [14]. An absolute loading calculated under one convention does not transfer to the other without the coefficient that belongs to it, which is why in-line traces are usually read as relative cycle-to-cycle integrals rather than as absolute substitutions.
Understand what that trace is and is not. It measures Fmoc groups removed in the current cycle, which corresponds to chains that were acylated in the previous one — a coupling readout by inference, and a clean one only when unreacted chains have been capped so that they cannot re-enter and carry an Fmoc group later. Its advantage over the bead test is not that it answers a different question but that it answers the same question with an integrated absorbance rather than a subjective colour, on every cycle, without consuming resin.
Acting on a positive test
A persistent positive can be a physical problem rather than a reagent problem. Larsen and Holm, synthesising leucine- and alanine-containing homo-oligopeptides by the Fmoc strategy, observed ineffective N-alpha-deprotection with piperidine in a sequence-dependent manner, found incomplete deprotection to be associated with subsequent slow or incomplete amino acid coupling, and concluded that difficult coupling and difficult deprotection in those systems share one physicochemical origin, beta-sheet formation [15]. A bead that keeps testing positive after a double coupling, once carried-over reagent has been excluded by washing, is consistent with the aggregation Larsen and Holm describe in their leucine and alanine systems, and repeating the acylation with fresh reagent will not fix that case.
The alternative response is to stop the failed chains. Montanari and Kumar's fluorous capping reagent tags all unreacted amines after the coupling step; the resulting fluoroalkylated amine can no longer couple in further steps and is stable to standard synthesis conditions, which is what allows the capped chains - which that paper calls deletion products - to be removed by flash fluorous chromatography [16]. Acetic anhydride capping does the same chemically without the purification handle. The trade is explicit: an uncapped failure propagates as a deletion peptide differing from the target by one internal residue, which is difficult to resolve; a capped failure stops as a truncated chain, easier to separate but still counted against yield.
Capping and monitoring interact in a way that is easy to get backwards. Once the unreacted amines have been tagged they are no longer free amines, so a ninhydrin test run after the capping step reads negative regardless of how the coupling actually went — the reagent has removed the very species the test detects [16]. Test first, cap second. A clean colour test taken after a cap is not evidence of anything except that the cap worked.
It is also worth being clear about the division of labour. Colour tests are process control: they decide, within a few minutes, whether to repeat a coupling before the next deprotection closes the window. They are not product analysis. Larsen and Holm's study of sequence-dependent deprotection failures used colour monitoring alongside FAB mass spectrometry and HPLC for product analysis [15], and that pairing is the right shape. Only cleaving a resin sample and running it tells you what you have built; the on-resin test tells you whether to keep building.
What the test licenses you to conclude
A strong blue on ninhydrin licenses one conclusion: a reacting primary amine was present in the tested sample. It does not tell you how much, does not tell you it was your chain end rather than carried-over reagent, and says nothing at all if your N-terminus is a secondary amine. A clean negative licenses less than most people assume — it places an upper bound on free primary amine at the visual threshold of the format, which is far above the level that determines whether a fifty-residue synthesis survives. Use the qualitative test as the cheap gate it is, use chloranil or an equivalent whenever the terminus is secondary, and put a quantitative measurement — ninhydrin in solution at 570 nm, or integrated Fmoc release — anywhere the answer has to be trusted.