Disclosure: I help Rosbacher and Römer Brunnen develop their markets in Taiwan. The historical SGS reports saved on this site serve as examples of how to read a report. Sample measurements, taste, and health claims are treated separately. This is not individual medical advice.

After my article on why Germans take water so seriously, friends asked: what do all those numbers on a bottle actually mean?

I understand the question. Before working with these two brands, I did not read mineral-water labels either. After months reading SGS Institut Fresenius reports, I realized that concentrations, quantitation limits, product limits, and calculated fields each have their own place in a table. Seeing a number is not the same as understanding it.

The original article tried to describe a water’s character through thirteen indicators, but also turned composition ratios into absorption claims. I now want to establish a reading order: first ask what was measured and in what units; then ask what question the number can answer.

Four-panel comic: Amount per liter; A 750 mL bottle does not make 233 mg/L the calcium amount in the whole bottle. Using the reported concentration, drinking 750 mL corresponds to 233 × 0.75 = 174.75 mg; This calculates the amount of a component consumed; it does not answer how much the intestine absorbs; Claims about effects need separate evidence; a label cannot stand in for a completed human trial. this was measured, this was calculated, and this claim still needs evidence.

First identify the product, sample, and test

The name Rosbacher on a bottle is not enough. The variety, packaging, sample, and report date must match. A different sparkling water’s pH or hardness cannot stand in for this one’s measurements.

Here are the two fixed reports used in this article:

ReportSample described in the reportDate
SGS 6886143Rosbacher Klassisch, 0.75 L glass bottle2024-05-21
SGS 4740147Bad Vilbeler Römer Brunnen, 1 L glass bottle2020-03-30

The second report identifies the sample category as Heilwasser. I use it to illustrate a laboratory table, not to turn a German category into therapeutic authorization in Taiwan. Neither report guarantees every batch on today’s shelves.

Next, read the column headings. Ergebnis means result, Bestimmungsgrenze is the quantitation limit, and Grenzwert is the listed limit. A second number beside a result is not necessarily another measurement, much less a daily intake recommendation.

For example, the Rosbacher report gives a fluoride result of 0.06 mg/L and a quantitation limit of 0.02 mg/L. These columns are not interchangeable. A value written as “< 0.005” must retain its less-than sign; it must not be copied as zero.

Download the full reports: Rosbacher SGS 6886143, Römer Brunnen SGS 4740147. All figures below refer to these two historical samples.

Amount per liter and amount consumed are two separate steps

Start with a few common entries. Every value in this table is in mg/L: milligrams per liter.

ComponentRosbacher sampleRömer Brunnen sample
Calcium Ca233550
Magnesium Mg111127
Sodium Na83.5649
Bicarbonate HCO₃⁻1,2362,849
Reported sum of dissolved minerals1,8404,912

A 750 mL bottle does not make 233 mg/L the calcium amount in the whole bottle. Using the reported concentration, drinking 750 mL corresponds to 233 × 0.75 = 174.75 mg of calcium.

Likewise, 250 mL of the Römer Brunnen sample corresponds to about 137.5 mg of calcium and 162.25 mg of sodium. This calculates the amount of a component consumed; it does not answer how much the intestine absorbs, whether the body needs it, or how much a particular person should drink.

This step already tells us something more useful than “high calcium” or “high mineral content”: the label gives a concentration, which must be multiplied by the actual volume consumed.

Read total content, hardness, and pH separately

Even a total needs its measurement method

TDS stands for total dissolved solids and is often expressed in mg/L. When comparing reports, check the original field and method too.

The same Römer Brunnen report lists a sum of dissolved minerals of 4,912 mg/L and residue at 180°C of 3,360 mg/L. These are separate fields in the report. We cannot pick a convenient number and treat every name for a total as interchangeable.

I have therefore removed the original article’s table placing different waters in a single supposedly “internationally customary classification.” A classification needs a source and purpose; a ranking needs comparable quantities.

Hardness is not total mineral content

USGS explains that hardness primarily concerns dissolved calcium and magnesium and is often expressed as mg/L as CaCO₃. This is a calcium-carbonate equivalent, not a simple addition of calcium and magnesium concentrations.

Higher TDS and higher hardness therefore do not mean the same thing. The original article also went too far in linking hardness directly to a “milky” mouthfeel and describing soft water as having only a watery taste. We can describe taste, but must leave room for individual perception rather than assign it all to one field.

pH is not a health score

pH describes water’s acidity or alkalinity; carbon dioxide and sampling conditions can affect it. USGS sampling guidance specifically notes that changes such as gas loss can alter a sample’s pH.

High bicarbonate alone does not establish a high pH. A bottle’s pH certainly cannot support promises to change a person’s “constitution” or treat disease. The table supplies no such evidence.

A calcium-to-magnesium ratio cannot calculate absorption

Dividing concentrations in the same units gives a calcium-to-magnesium mass ratio of about 233 ÷ 111 = 2.10:1 for the Rosbacher sample and 550 ÷ 127 = 4.33:1 for the Römer Brunnen sample.

These ratios describe sample composition. They do not measure human absorption.

The original article called a ratio near 2:1 “ideal” and said departures reduced absorption. The two composition reports cannot support that conclusion. An attractive ratio is no reason to give a water an absorption-benefit check mark.

NIH ODS information on calcium explains that absorption depends on factors including intake, age, and food sources. Assessing a person’s nutrition requires their diet and circumstances, not just division of two numbers on one bottle.

Mineral water can provide a calculable amount of a component. Claims about effects need separate evidence; a label cannot stand in for a completed human trial.

Trace elements are clues, not ready-made health benefits

Lithium, strontium, silicon, and boron can give a water an intriguing story. But detection only answers questions about presence and concentration, not the health effects of drinking it.

Even similarly named fields cannot always be added. The Rosbacher report lists silicon Si at 7.5 mg/L, silicon dioxide SiO₂ at 16.1 mg/L, and silicic acid H₂SiO₃ at 20.8 mg/L. These involve different expressions of the same element; they are not three unrelated nutrients to add together.

Limits must retain their original purpose. The 1.5 mg/L fluoride value in WHO drinking-water guidance is a guideline value. Combining it with a lower number does not create an “optimal intake range” everyone should seek. The document also calls for consideration of water consumption and other fluoride sources. WHO fluoride fact sheet

In another article, I distinguish lithium research populations and formulations from mineral water. Composition analysis can guide further questions; it cannot establish preventive or therapeutic benefits for a particular water.

My reading order next time I pick up a bottle

First check the variety and date, then the per-liter units, then calculate the amount actually consumed. For total content, hardness, and pH, find the relevant field and method separately. For attractive ratio or trace-element claims, ask whether the evidence measures composition or effects in people.

I still like thinking of a composition table as a source’s geological calling card. But the card cannot tell us the age of the strata, the water’s entire journey, or its effects on people. Those questions need other evidence.

I began reading reports to help friends understand a bottle of water. Now I care more about whether readers can say for themselves: this was measured, this was calculated, and this claim still needs evidence.