Prepared by: Brandon LaBumbard, microbiologist at Arcaea

Introduction: A question that fell from the sky

On February 23, 2026, a nor’easter dropped 17.1 inches of snow on Boston. The next morning we scooped and packed about four liters of snow off the balcony of the Arcaea office and asked a simple question: what’s alive in fresh snow, and what possible scents are in there?

Snow isn’t just frozen water. A snowflake begins when water vapor freezes around a tiny particle in a cloud, like a speck of dust, a bit of pollen, or a microbe. Then, as heavy snow falls, it sweeps even more particles out of the air. Every flake is a little archive of the sky it passed through.

This report follows that snow from our balcony to the lab, into the genomes of the microbes it carried, and finally to a perfumer's brief. We called it Project Elsa (of course). 

What we did, in four steps

From collection in late February, we went from four liters of snow to a list of 119 molecules with a known smell by early June. Each step gave us a closer look of what was possible within the snow microbial world and scent production. We also relied on public databases that anyone can check to identify the scent molecules.

1. Collection: We melted four liters of snow into about one liter of water and concentrated the cells in two ways: spinning them down in a centrifuge, then filtering what stayed suspended through a membrane.

2. Grow: We added the samples onto nutrient plates and transferred new microbial colonies onto fresh plates until each was pure. We also kept a sample of the raw snow water to sequence its DNA directly, so we could see the microbes that would not grow on a plate.

3. Read: We sequenced the full genome of each microbe and the DNA from the raw snow sample. Think of it as being handed the instruction manual of how the microbes work and make things. Specialized programs were then used to find and identify the genes and matched each one to a known enzyme in a global reference library of biochemistry database (KEGG).

4. Translate: An enzyme carries out a known reaction, and a reaction makes known products. Using bioinformatics, we listed every product these genomes could make, then checked each against five curated smell databases used by flavor and fragrance chemists. Molecules with a recorded odor made the cut.
   
   We also ran the same analysis on the published genome of Glaciibacter psychrotolerans, a microbe adapted to snow and glaciers, as a point of comparison. 
   

Table 1. Overview of sequenced data that was processed to determine what pleasant scents were possible in the genomes of microbes isolated from snow. Each step keeps only what the step before can support, which is why tens of thousands of genes end in a short list of scent molecules.

Microbes, who are they?

Five microbial species grew from the snow samples we collected: three fungi (a yeast and two molds) and two bacteria. Their genomes read like toolkits for thriving on a bright, cold surface, which was very exciting to see.

A snowpack is a demanding place to live. Snow reflects 80 to 90% of sunlight, so anything living on it is lit from above and below. Cold water also holds extra oxygen, which, combined with all that light, puts constant stress on cells. Cell membranes must also stay supple below freezing temperatures. This means that these microbes have evolved to live in some pretty hardy conditions. Here's what we found:

Table 2. Microbial species cultured from snow and what the sequencing data revealed about genes present in the genomes based on KEGG orthology.

Trickster Microbes

Two of the microbes were difficult to separate at the start to get pure cultures. The pink yeast, Cystobasidium ritchiei, and the bacterium Pseudonocardia sediminis kept growing together on the same plates. We picked colonies, plated and replated them multiple times before we had pure cultures we could sequence.

We suspect they were living together in the snow, possibly in a shared biofilm: a community of cells held together in a film they build around themselves. Microbes in biofilms often trade nutrients and protection, which would explain why these two held on so tightly. We didn't test this directly, so for now it's our best hypothesis, and makes for a nice story.

These two microbes also were trying to catfish us. Based on looks alone, the colony we took for the yeast (fungus) turned out to be the bacterium, and the one we took for the bacterium turned out to be the yeast. Only after sequencing did we realize. Pseudonocardia sediminis belongs to Actinobacteria, a group whose colonies can look more like fungi than bacteria with some “fuzzy” looking crusts, while Cystobasidium ritchiei grows in smooth, glossy colonies that can pass for bacteria. It was a reminder that in biology, appearance is a hypothesis and the genome can help give a clearer answer.

The microbes in the DNA

Sequencing DNA straight from the snow water showed a different crowd: mostly bacteria from a family (Oxalobacteraceae) that was very different from what grew on our plates. Much of the snow's life never makes it into a petri dish, which is well documented in other systems as well! More than 99% of the microbial life in the ocean is thought to be unculturable and so it makes sense that we were not able to culture everything from the snow samples. But, this DNA sample alone revealed 89 of the 180 scent molecules identified, 9 of them found nowhere in the microbial data from our cultured microbes.

Microbes that help make ice

Some microbes carry ice-nucleating genes, meaning they help form ice crystals. We were curious whether we'd find any in our sample. We didn't, but we did find two other gene families that help microbes survive in the cold: ice-binding proteins, which help cells withstand being frozen, and cold-shock proteins, which help them stay metabolically active at low temperatures. Seven genes across three of the microbes and the DNA sample carried an ice-binding domain, and twenty for a cold-shock domain. These genes are not linked to specific reactions, so they don't feed into the scent list, and also shows that more is waiting to be described.

Color that leads to scent

This was very interesting! One finding links the snow's pink pigment directly to its scent chemistry. The pink yeast makes torulene, a carotenoid. Penicillium nordicum carries an enzyme, torulene oxygenase, that cleaves torulene. One of the pieces is 3-methyl-2-butenal, a small volatile molecule with a sweet, nutty smell. In other words, the same chemistry that shields a microbe from snow glare can also release a scent.

Pigment and perfume can sit on the same pathway, which is why bringing the color into the fragrance felt important. Snow Beginnings is tinted a soft pink, and its brightness was briefed as glimmers on fresh snow at a pink dusk that just happens to also resemble the pigments in the microbes.

The scent molecules

Finally, we are at the scent molecules. The data points to 180 molecules with a recorded smell from the reference databases. Of these, 119 come from our internal sequencing data and did not appear in our earlier Skyland work on clouds (Cloud Reverie and rain (Wandering Rain). Among these, 46 were considered pleasant snow-specific molecules with leading fruity, floral, and balsamic notes. A selection of which are detailed in Table 3.

Table 3. Scent molecules from the snow microbial genomes that pair with the notes of Snow Beginnings. Scent descriptions come from five curated odor databases: Flavornet, The Good Scents Company, FlavorDB, AromaDB and Leffingwell.

Translating the science to perfume

The molecule list became a creative brief and in mid June, we handed it to perfumer Michael Nordstrand. Snow Beginnings is the third chapter of our Skyland Collection, which explores the nature above us. It joins Cloud Reverie (developed by perfumer Daniela Andrier) and Wandering Rain (also developed by Michael Nordstrand). Michael built with perfumery materials, and the snow data shaped his palette rather than dictating it (Table 4). 

The idea at its center is overwintering, the ecological term for how living things persist through winter. It isn't inactivity, it's a different part of the life cycle. So the fragrance does not try to recreate how winter smells, it captures what winter makes possible.

Table 4. A few highlights of Michael Nordstrand’s note compositions for Snow Beginnings. Top Notes: Japanese Mint, Mistletoe, Pink Grapefruit; Middle Notes: Moon Pear Saké, Crystallised Rose, Lychee; Base Notes: White Cypress Absolute, Musk, Deer Moss

What this does and doesn't tell us

We understand snow a little better now, and there's plenty left to explore. We want you to know exactly how far these results go.

• This is a prediction, not an exact replica. We read what these microbial genes can make, we did not capture the air above the snow and measure its molecules. A gene for an enzyme shows capability, not activity. The molecules in Table 3 take part in enzymatic reactions, though not always as the final product and these molecules are not always made. Those reactions run in both directions, depending on what substrate is available at the time.

• This was also a single snowfall in one city. Another storm in another place or time could carry different microbes. We are asked this question about our work on Cloud and Rain, and we very much believe that different locations, seasons, and conditions would create different results. Still, many of the isolates from snow are cold-adapted, so it's encouraging that our data may relate to other snow events.

• We grew the microbes at room temperature on a general-purpose medium. Microbes that prefer the cold, or need special food could have been missed. That's why we also sampled DNA straight from the snow, to get a more complete idea of the microbes found in the snow.

• Databases have gaps and are always being updated with new information. Many molecules these snow microbes make may not be catalogued yet. Enzyme and odor databases are built from decades of published work, but they're incomplete. The cold-adaptation genes above are one example: they have no known reaction, and hopefully one day will be mapped out.

• A molecule's described scent comes from experts smelling it pure and at set strengths. In a living environment, or in a perfume, it behaves differently so our odor profiles may smell different depending on many factors.

Glossary

Always happy to expand your vocabulary and make this a jargon-free environment.

• Biofilm. A community of microbes living together inside a protective film they produce.

• Carotenoid. A family of mostly red, orange and yellow pigments, including the beta-carotene in carrots. Microbes can use them as sunscreen and antioxidants.

• Enzyme. A protein that speeds up one specific chemical reaction. Genes are instructions for building enzymes.

• Genome. An organism's full set of DNA.

• KEGG. The Kyoto Encyclopedia of Genes and Genomes, a public reference library linking genes to enzymes, reactions and molecules.

• Volatile. A molecule light enough to drift into the air, which is how we smell it.