MAI-Alchemy · Analytical Chemistry, Taught by Practice
The Analytical Column
where ideas resolve54 pieces · 10 subjectsFoundations · Analytically Speaking
Analytical chemistry for the people who do it: short, sourced lessons on how a result is really made, from the sample to the people in the lab, and essays from the bench. Free to read, no account needed. Start here →
Showing ·
Read like a student: select any passage (or tap the ✎ pen) to put it in your own words, research it across the site, share it, or ask me about it. Every underlined term opens a plain-language card. New here? Start here → · Glossary
Reading where there's no signal? Pick all or just the pieces you want; they stay on this device.
Foundations
Foundations for the working analyst — each piece grounded in the literature and checked before it ran.
- Reading concentration from a single numberA binary solution hides its composition in its density. Why that one number is enough — and where it stops working.
- The power of the densitometerDensity is one number, measured in a minute. This is what it does on its own, and what it does when you give it a second measurement to work with.
- When one number isn't enoughDensity can name one unknown. Add refractive index and you can name two. Here is how the pair resolves a mixture neither could resolve alone.
- Fingerprinting a fluidDensity says how heavy a fluid is. Viscosity says how it flows. Measure both and you stop weighing a fluid and start recognising it.
- The oscillating U-tube, in depthA modern density meter turns a frequency into a density with one equation and two constants. Here is how a vibrating glass tube does it.
- The second number your density meter already measuresMany density meters also clock how fast sound travels through the sample. That number is an independent handle on composition — most useful where density struggles.
- Calculated, not measuredSome of the numbers on a fuel report were never measured. They are computed from density and a distillation curve — and density is the load-bearing input.
- Viscosity index, and what it really tells youEvery liquid thins as it warms. Viscosity index is a single number for how fast it thins — and for what that number leaves out.
- When one viscosity number isn’t enoughFor water, viscosity is one number. For paint, grease, and blood, it depends on how hard you push — and one number becomes a trap.
- Reading wear from bulk propertiesYou do not have to name every molecule in a used oil to know it is in trouble. A drift in cheap bulk properties, from a known baseline, is the early warning.
- The number the instrument can't give youAn instrument reports a response — an area, an absorbance, a current. Never a concentration. Building the curve that maps one to the other is where you actually learn the analysis — if you read what it's telling you.
- Calibrate inside the sampleWhen the sample itself bends the response and you cannot rebuild that matrix in a standard, you move the calibration to where the problem lives — and pay for it in a way worth respecting.
- The reference that rides alongYou rarely inject exactly the same volume twice, or lose exactly the same amount in prep. So you stop trusting the raw signal and trust a ratio — the analyte measured against a companion that suffers what it suffers.
- Where the signal loses to the noiseEvery method has a floor: a level below which you cannot say how much analyte is present, and a lower one below which you cannot even say it is present. Both are set by two things together — how much the blank wanders, and how steeply the method responds.
- Where the line stops being straightA calibration is only good where the response actually tracks concentration. Above that, the curve bends, the top standard falls off the line, and a straight fit forced through it distorts the whole calibration. So you prove where linearity ends — you do not assume it.
- What calibration isNo two instruments are identical, and they do not need to be. Calibration ties each instrument’s signal to the same standards, so the reference, not the hardware, decides the number.
- The detector is the constantOf all the hardware, the detector is the most tightly specified and tested part of an instrument. Overlay one method’s calibrations and the shape holds while the level moves, so when calibration is lost, look upstream first.
- Gravimetry under the standardsA calibration curve is only as true as the standards behind it. Weigh what you put in, correct for what it really is, and calculate the concentration you actually made.
- Is the calibration still holding? An internal standard plus a CRM as your QCAn internal standard keeps each run steady; a certified reference material tells you whether the steady answer is still right. Chart one, test the other, and watch both halves of accuracy.
- A calibration curve is a team artifactOne analyst’s tight replicates describe one analyst. The lab’s real precision is the spread across people, days and instruments, and the curve every result depends on belongs to all of them.
- Calibration-free? Reference methods and normalisation, honestlyEvery method that skips the fitted curve trades it for an assumption. Normalisation, reference methods and pre-loaded factors each have one; name it, check it, and keep calibrating the instrument.
- How to read a QC chartA QC chart is not a verdict on randomness. Its patterns are questions about what changed, and in an injection method the first answer is usually how the sample got in.
- The ideal gas law at the instrumentPV = nRT runs through every gas chromatograph: how much sample a gas loop delivers, whether a liquid injection fits its liner, and how the carrier flows.
- What a titration actually measuresA titration counts. A measured amount of one substance reacts with another, and the balanced reaction turns the volume into an amount.
- Standardising the titrantA titrant’s label is not its concentration. Standardise it against a primary standard, under the conditions you will use it, and keep its factor on a chart.
- Finding the endpointEvery way of finding the endpoint looks for the steepest point of the curve: an indicator whose colour change sits inside the jump, or an electrode and its derivative.
- The burette and the dropOnce the chemistry is right, most of the error left is in the volume: the glass, the reading, the last drop and the temperature.
- Acid and base number: the methodsAn acid or base number counts whatever reacts under one method’s conditions, expressed as potassium hydroxide. The number belongs to the method.
- Reading an oil's acid and base numbersA used oil’s numbers mean something against its own new-oil baseline and its own trend, measured one way from the first sample to the last.
- Karl Fischer: titrating waterKarl Fischer titration counts water through iodine, one for one; the cell, the drift and the sample’s own reactions decide whether the count is right.
- Redox and complexometric titrationsRedox and complexometric titrations count the same way: write the ratio from electrons or from EDTA’s one-to-one complex, and set the conditions the method states.
- Quality control for titrationQuality control checks what a titration depends on: the blank, the titrant, a line through zero, and a check sample that moves only when something has changed.
- Units and dimensional analysisWrite the unit with every number, let the units cancel, and look at what is left: a wrong unit at the end is a wrong setup, caught before it becomes a wrong result.
- Concentration: the ways labs say how muchEvery concentration is a fraction. Know what it divides by, make the solution so the label is true, and dilute by keeping the amount constant.
- Chemical calculations: moles, mass and the balanced equationGrams to moles with the molar mass, moles to moles with the balanced equation, and back: the chain behind every solution, and why the first number, what is in the bottle, decides it.
- GC quantitation: external and internal standards, by the numbersEvery reported number is an area turned into an amount. Work the arithmetic of external and internal standards, by volume and by mass, and check any result your software gives you.
- Standard addition in GC, and mass versus volume resultsCalibrate inside the sample when the matrix changes the response, keep the units of the x-axis honest, and convert mass percent to volume percent with two densities at one temperature.
- Sample weight or total weight? Checking the software's internal standard mathHalf the labs typed the gasoline's weight, half the total. Work D4815's calculation from first principles, size the error, prove which entry your setup needs, and lock it down.
Analytically Speaking
Essays first published on LinkedIn, 2024–2025 — thirty years of lab life, in my own words.
- Why Procedures Don’t Matter
- Why Winning the Game Doesn’t Matter
- Why Failing Matters
- Why Most LIMS are Frustrating
- Why You Matter
- Why the Sample Point Matters
- Why Sample Quality Matters
- Why Sample Introduction Matters
- Why Challenges Matter
- Why Temperature Matters
- Why The Internal Standard Matters
- Why Variability Matters
- Why The Chromatogram Matters
- Why Trying Matters
- Same software, same method, two answers
- Why Alchemy Matters