Your scale decides how many materials your formula can have — on a 30 mg scale, 14 of 17 cannot be weighed

Someone about to make their first perfume posted an equipment list and a formula, and asked four questions. The third is this piece's subject:

"Is my scale (30 mg resolution) good enough to start? I know it's far from ideal, but is it good enough for learning, or will it cause too many problems?"

This can be worked out, and the answer is specific.

The short version

• The weighing rule of thumb: for 1% accuracy, each weighing should be at least 100 times the readability. A 30 mg scale means at least 3 g every time.
• Their formula is 20 g, 17 materials. Fourteen of them fall below 3 g — fourteen out of seventeen cannot be weighed to 1% accuracy.
• Relax to 5% accuracy (20 × readability = 600 mg) and ten still fall short.
• The extreme case is cinnamon bark oil: 0.2% = 40 mg, which is 1.3 scale divisions.
• Swap in a 1 mg scale — same formula, same batch — and the count of materials failing 1% accuracy drops from 14 to 1. That scale costs under twenty dollars online.
• And smaller batches make it worse. The forum advised a 1 g trial; on a 30 mg scale, a 1 g batch leaves all 17 unweighable.
• This is not a peculiarity of their formula. Across 954 public formulas, 24.6% of material entries sit below 1% and 51.3% below 3%; the median formula has 7 materials below 3%.
• So dilutions are not an advanced technique. They are what the scale forces on you. And they cost something: serial dilution methods are particularly prone to error propagation (Walling et al. 2011).

About 9 minutes.

What the rule is

A digital scale's specification carries two numbers people confuse.

• Readability (d): the smallest displayed increment. Theirs is 30 mg.
• Minimum weighable mass: not a specification but something you calculate from the accuracy you want.

The common rule of thumb:

Relative accuracy wanted	Minimum weighing
1%	100 × d
5%	20 × d

The reasoning is direct. If the smallest increment is 30 mg, any single reading carries at least that much uncertainty. Weighing 3 g, 30 mg is 1%. Weighing 300 mg, it is 10%. Weighing 40 mg, 30 mg is 75%.

(To be clear: 100×d is a rule of thumb, not a regulation. The real minimum also depends on the balance's repeatability and linearity and on draughts, vibration and temperature where it stands. Limits are at the end.)

Their formula, line by line

20 g batch, 30 mg scale. "Divisions" is the mass divided by 30 mg.

	Material	Dose	Mass	Divisions
✓	Ambrox Super	24.0%	4,800 mg	160.0
✓	Hedione HC	20.0%	4,000 mg	133.3
✓	Bergamot FCF	19.0%	3,800 mg	126.7
△	dihydro-β-ionone	8.0%	1,600 mg	53.3
△	lemon oil FCF	6.0%	1,200 mg	40.0
△	ambrettolide	5.0%	1,000 mg	33.3
△	Lyfral	3.6%	720 mg	24.0
✗	sweet orange oil	2.5%	500 mg	16.7
✗	Muscenone Delta	2.2%	440 mg	14.7
✗	Exaltenone	2.2%	440 mg	14.7
✗	guaiacwood oil	1.5%	300 mg	10.0
✗	ginger CO₂	1.5%	300 mg	10.0
✗	black tea accord	1.5%	300 mg	10.0
✗	Muscone Laevo	1.0%	200 mg	6.7
✗	neroli accord	1.0%	200 mg	6.7
✗	frankincense oil	0.8%	160 mg	5.3
✗	cinnamon bark oil	0.2%	40 mg	1.3

✓ = above 100 divisions (1% accuracy): 3 materials. △ = 20 to 100 divisions (5% accuracy): 4 materials. ✗ = under 20 divisions: 10 materials.

Three of seventeen can be weighed properly.

The cinnamon line deserves separating out: 40 mg is 1.3 divisions. That scale will display 30 mg or 60 mg, and the 40 mg you want does not exist on it.

A reply in the thread ran the same calculation and gave the right fix:

"20g is way too large of a batch, especially for your very first trial. Use dilutions and do 1g… So for example you'll need 2mg of cinnamon bark. You'll want to make a 5% dilution and use 40mg (about 3-4 drops) or a 2% dilution and use 100mg… A 0.03g scale won't be sufficient."

That person did the arithmetic correctly.

What changes if you change the scale

Same formula, same 20 g batch, only the scale changes:

Scale	Needed for 1%	Materials failing	Needed for 5%	Materials failing
30 mg	3,000 mg	14 / 17	600 mg	10 / 17
1 mg	100 mg	1 / 17	20 mg	0 / 17

Fourteen becomes one.

The one that still fails is cinnamon bark oil at 40 mg — and 40 mg on a 1 mg scale is 40 divisions, comfortably inside the 5% band. For a first perfume that is entirely acceptable.

Somebody in the thread named the price directly:

"You can buy a 0.001g scale on amazon for under $20 too. A 0.03g scale won't be sufficient."

That is the most practical sentence in this piece. Buy that scale before the seventeenth material.

Smaller batches make it worse

This is counterintuitive, and it puts the forum's advice in conflict with their scale.

Same 30 mg scale, varying the batch:

Batch	Failing 1%	Failing 5%
20 g	14 / 17	10 / 17
10 g	17 / 17	12 / 17
5 g	17 / 17	14 / 17
1 g	17 / 17	17 / 17

At a 1 g batch, that scale cannot weigh a single one of them.

So "make small batches" and "use a cheap scale" cannot both hold. Small batches are the right advice — trials should be small — and they require a better scale, or more dilutions.

This is not a peculiarity of their formula

Laying out the doses across all 954 public formulas, solvents excluded, gives 16,977 material entries:

Dose	Entries	Share
< 10%	13,897	81.9%
< 5%	10,965	64.6%
< 3%	8,703	51.3%
< 1%	4,184	24.6%
< 0.5%	2,204	13.0%
< 0.1%	218	1.3%

More than half of all material entries sit below 3%, and a quarter below 1%.

And the median formula looks like this:

Threshold	Materials below it, median formula
< 5%	10
< 3%	7
< 1%	2

A typical formula has seven materials dosed under 3%.

In a 20 g batch, 3% is 600 mg. On a 30 mg scale that is 20 divisions — exactly on the boundary of 5% accuracy.

In other words: if you are making normal formulas, you will keep hitting this.

So dilutions are not optional

This is the one thing to take away.

Many people treat dilutions as an advanced technique, or as a convenience for evaluation. The arithmetic above says they are a necessary step forced by the physical limits of the scale.

You do not dilute for convenience. You dilute because you cannot weigh.

The mechanics are direct:

• A material dosed at 0.2% is 40 mg in a 20 g batch.
• Make a 10% dilution and that 40 mg becomes 400 mg of solution — 400 divisions on a 1 mg scale, 13 on a 30 mg one.
• Make a 1% dilution and it becomes 4,000 mg — enough even for the 30 mg scale.

The cost is that solvent enters the finished product. Somebody in the thread flagged it:

"Dilute in alcohol when possible so you don't end up with too much tec or dpg in the final result."

That is right, and the constraint is real. With five materials each coming in as 10% solutions in DPG, that DPG accumulates.

But dilutions carry their own error

Before recommending them, their cost should be stated.

Walling and colleagues were addressing quality control of serial dilutions in drug screening in the Journal of Laboratory Automation in 2011, and their problem statement is explicit:

In general, serial dilution methods are particularly prone to error propagation because each dilution is dependent on the previous concentration.

So a longer dilution chain means larger error.

In practice:

• Do it in one step where you can. Going straight from neat to 1% beats 10% then 1% — unless the first step cannot be weighed (as with dimethyl sulfide).
• Label every bottle with what it is. The person in the previous piece was tripped up by a dilution of a dilution.
• Weigh the dilution rather than counting drops. The thread's "40mg (about 3-4 drops)" is a useful conversion, and drop size depends on viscosity and pipette bore.

Professionals get this wrong too

The last section exists to place expectations properly.

Shah and colleagues did something direct in the World Journal of Gastroenterology in 2013: they had 12 prescriptions for 2% diltiazem cream filled at 12 different retail pharmacies openly offering compounding services, with two refills allowed each, producing 36 preparations, then measured content by HPLC.

The USP standard for potency is 90%-115% of label claim.

The results:

Of the 36 preparations, 5 (13.89%) were suprapotent, ranging from 117.2% to 128.5% of label claim, and 13 (36.11%) were subpotent, ranging from 34.8% to 89.8%. Fourteen (38.9%) lacked content uniformity according to the USP standard. Nine of the 12 pharmacies (75%) failed USP potency or content-uniformity specifications for at least one of the three fills.

The lowest was 34.8% of label claim — a third.

These are licensed professionals with equipment and training.

The limits of this comparison need stating: a cream's problems include mixing and homogeneity, not only weighing, and these are US retail pharmacies rather than perfumery. I cite it not to claim perfumery does the same, but to show that content error in small-scale hand compounding is a measured and substantial problem.

If professionals working to a published standard fail at that rate, then a beginner weighing a 17-material formula on a 30 mg scale is not making the formula they wrote down.

Which does not mean they cannot make something good. It means they cannot reproduce it.

A minimum equipment list

Collecting the three commenters' advice, marking which items the arithmetic actually requires:

Buy now:

• A 0.001 g (1 mg) scale, 50-100 g capacity. This is the only item the arithmetic demands.
• Small glass vials: a batch of 1-4 mL for dilutions, a few 15 mL, a couple of 30 mL.
• Disposable pipettes. One commenter specifically likes the 0.2 mL ones.
• Alcohol: SDA 40B 95% or higher. (The poster has 96.4% drinking Weingeist — the strength is fine, and the difference between drinking spirit and perfumer's alcohol lies in denaturants and the regulations that follow, which I have not checked.)
• Blotters. One commenter notes they double as a lab scoop.
• Labels and a notebook.

Useful but can wait:

• Glass pipettes, to wash and reuse
• Isopropyl alcohol for cleaning and de-scenting
• A stainless steel tray — one commenter says you will knock something over, and a tray at least contains it
• Amber bottles, if things sit out in the light

Skip for now:

• Beakers, glass stirring rods, funnels, syringes — at 1-20 g you do not need a beaker.

What this doesn't establish

100×d and 20×d are laboratory rules of thumb, not regulations. Sources differ (some use 50×d), and the real minimum weighing also depends on repeatability, linearity, draughts and vibration. I use these thresholds to quantify the problem, not because they are a standard.

All I know about their scale is "0.03 g / 30 mg resolution". I treat that as readability. If it is repeatability or a stated minimum weight, the numbers change.

The formula is as the poster wrote it and I did not audit its provenance. Someone in the thread asked whether the percentages came from AI, and the poster did not answer.

The corpus dose distribution excludes solvents, and corpus percentages are shares of a concentrate. That is the same level as their 20 g of "oil", so the comparison holds.

Shah's study is creams, not perfume. Content uniformity in a cream owes a great deal to mixing and phase separation, which is not weighing error. It describes US retail pharmacies in 2013. I cite it for the general fact that content error in small-scale hand compounding has been measured and is substantial.

Walling's paper is automated dilution QC for DMSO systems, using 96- and 384-well plates and a fluorescent dye. I cite only its statement about error propagation in serial dilution, which is a general remark rather than their experimental result.

I did not check the alcohol point. The difference between drinking spirit and denatured perfumer's alcohol — denaturants, duty, regulation — varies by country, and I did not look up the German position.

I offer no opinion on how the formula smells. Commenters gave formulation advice (black tea too high, cinnamon too high, swap Exaltenone for Exaltone); those are their judgements and I am not in a position to assess them.

References

M. Shah, L. Sandler, V. Rai, C. Sharma, L. Raghavan, Quality of compounded topical 2% diltiazem hydrochloride formulations for anal fissure, World Journal of Gastroenterology, 19(34), 5645-5650 (2013). PMID 24039356. doi:10.3748/wjg.v19.i34.5645

L. A. Walling, N. R. Peters, E. J. Horn, R. W. King, An inline QC method for determining serial dilution performance of DMSO-based systems, Journal of Laboratory Automation, 16(3), 216-221 (2011). PMID 21609707. doi:10.1016/j.jala.2011.01.001

Related: weighing and dilution, I can't smell it, which trials to keep, solvents.