Tomatoes Worth Keeping

The seed catalog offers a black tomato, a striped tomato, a tomato shaped like a pepper, and one whose description makes it sound like a minor royal appointment. Your garden offers six spaces.

Our worked garden has six spaces. That is enough to make the choices visible, though too few for many breeding and preservation projects. A person with room for sixty plants can postpone some decisions until the seedlings are already growing. With six, every interesting addition displaces something else. The early red tomato may be ordinary, but it supplies the kitchen while the spectacular late one is still considering its options.

Before buying anything, decide which of three jobs you want the garden to do. You might want a more varied harvest. You might want to keep a named variety going from seed. Or you might want to cross two plants and select something new from their descendants. These projects overlap, but they require different records and different uses of space.

For a varied harvest, the six spaces can hold six varieties. You can compare colors, sizes, ripening times, and uses in the kitchen. You do not have to breed anything. There is no lesser achievement in finding a collection that gives you tomatoes you enjoy through more of the season.

For seed preservation, the important result is a dependable next generation of the variety you already value. You need to know which plant supplied a packet, how its flowers were pollinated, and whether it matched the variety you meant to keep. A surprising offspring may be interesting, but surprise was not the purpose of that packet.

For breeding, some spaces become experimental. The offspring may disappoint you. If all six plants belong to a breeding project, you have made the summer's tomato supply depend on an experiment. That may be acceptable. Decide in spring, before the empty cutting board in August.

Give each space a job

Here is a possible allocation for the first year. The garden and its results throughout this book are hypothetical.

This plan sacrifices the thrill of six different varieties. In return, it gives you a chance to see whether two plants of the same parent behave alike and to attempt more than one cross. It also protects two spaces for ordinary eating.

The following year needs its own plan. Seeds from a successful cross will produce a first generation. Seeds saved after those plants self-pollinate will produce the next, more variable generation. The cross therefore creates a future demand for space. A packet can contain far more possibilities than a garden can test.

Write that demand down before making several crosses. “I will grow four offspring from one cross” is modest but understandable. “I will see what comes up” leaves the eventual choice to the number of pots you happen to find. You may still get an enjoyable tomato, but you will know less about what you have sampled.

Choose a problem you actually have

A breeding goal should describe a change you would notice. Perhaps the tomatoes you like ripen too late for your site. Perhaps the plants outgrow their supports. Perhaps you want small fruit that you can eat whole, but the variety you tried split too often in your conditions. Start with that problem.

“Better flavor, higher yield, earlier, smaller plant, and resistance to everything” is a list of wishes. It gives you no way to decide between two offspring, each of which improves some things and worsens others. A more workable first goal is one main improvement and a short list of qualities you are unwilling to lose.

For example: seek earlier ripe fruit while keeping a flavor you enjoy. Record ripening dates and repeat the tasting. Treat plant size and usable yield as constraints. If the earlier plant produces so little that it no longer serves the kitchen, the date alone has not solved your problem.

The garden itself must be part of the description. A plant next to a warm wall has not necessarily inherited greater earliness than one in a cooler corner. A container that repeatedly dries out is a different growing environment from a deep bed. Sketch the site and mark those differences.

Use the first season to learn the parents well enough to choose deliberately. If neither suits you, there is no obligation to cross them merely because the seed packet has arrived. Buying a better-suited variety next year may make more sense than spending several generations trying to repair a poor starting choice.

This book assumes ordinary cultivated garden tomatoes and a small project carried out with hand tools. It does not promise a new named variety by autumn. It offers a way to make a cross, keep track of what follows, and decide whether the result deserves another place in the garden.

A variety name is useful because it allows a gardener to ask for something again. It is not the plant's whole history. Two packets with different names may offer similar things; two plants with the same name may not perform identically when their seed sources or growing conditions differ.

Keep the original packet or photograph both sides. Record the supplier and the year. A bare label saying “yellow tomato” is enough for a rough garden map. It is poor evidence when you later want to know whether an unusual offspring came from a deliberate cross, an accidental mixture, or a plant whose name you never knew.

The word cultivar refers to a cultivated variety. In this book, variety is the convenient everyday term for a recognizable cultivated tomato. We will use parent, family, and line more narrowly when following a breeding project. A parent is a plant used in a cross. A family is a group of descendants whose recorded origin they share. A line is material being maintained or selected through generations. Writing these words consistently makes the notebook easier to read.

Open-pollinated is not a synonym for wild

An open-pollinated variety can be maintained by suitable seed-saving practices so that its offspring remain broadly like the variety. Many such tomatoes have histories of deliberate selection. The fact that a gardener can save their seed does not mean that nobody bred them.

Heirloom is a historical or cultural description, used with varying definitions. It is not a separate mechanism of reproduction. An older variety can be a valuable starting point; age by itself tells you little about its performance at your address. A recently selected open-pollinated variety may suit you better.

A commercial F1 hybrid is usually produced by crossing selected parent lines. Seeds collected from a tomato on that F1 plant can germinate, but the resulting generation need not reproduce the F1's combination of characteristics. The University of Minnesota's seed-saving guide makes this distinction explicitly: hybrid tomato seed is not inherently sterile, and viable offspring are not the same thing as predictable offspring.

That distinction prevents two opposite mistakes. You need not throw away an interesting breeding idea because one parent is a hybrid. Nor should you promise a neighbor that seed saved from a purchased hybrid will reliably recreate the plant on the packet. Your purpose determines whether variation is welcome.

Preserve the name, or start a new record

Suppose you buy an open-pollinated variety called A. You grow it, protect flowers for self-pollination, and save seed from plants that fit its description. Your aim is to maintain A.

Now suppose you deliberately cross one of those plants with variety B. The seeds from that cross should not go into an envelope labeled only A, even though the fruit grew on the A plant. Give the cross its own record. Later chapters explain why the fruit and the embryos inside its seeds have different roles in the story.

A third possibility is an unexpected plant among what you believed to be A. Perhaps it is useful. Keep its identity separate while you investigate. “A, unexpected plant 3” is an honest temporary label. Immediately renaming all its descendants as a new stable variety claims more than one observation can establish.

There is room for affection in a notebook. A nickname can make a plant memorable. Keep a plain identifier beside it. The nickname “the one by the broken pot” stops working after you replace the pot; a plant number continues to identify the material.

Choose differences that matter at harvest

Color is an easy difference to shop for because a photograph shows it quickly. Other differences may matter more to your kitchen. A tomato that ripens when you are away is poorly timed, however handsome it looks. A large crop that requires more processing than you can manage may become a chore.

List the uses you actually have. Eating while working in the garden, slicing for sandwiches, cooking, and giving fruit away place different demands on the crop. You need not grow a specialized variety for each use, but you should know why a candidate deserves one of the six spaces.

Growth habit also affects the practical plan. A determinate plant and an indeterminate plant organize growth and fruiting differently. Read the variety's description rather than treating either word as an exact height specification. Supports, container suitability, and the expected harvest pattern still depend on the particular cultivar and conditions. Local guidance, such as Minnesota's home-tomato guide, is a starting point for matching the plant to the season; transplant dates from one region should not be copied into another.

Make a short comparison sheet before ordering. Give each candidate a reason, a known limitation, and a question to test. “Good reputation” is less useful than “I want to compare its first ripe date with the tomato I already grow.” The latter can still teach you something if the new arrival loses.

The sheet only needs enough detail to preserve your intentions. When the crop starts ripening, you should be able to recover the question you meant to ask in spring.

A tomato flower is small enough to make a first crossing attempt feel like repair work on a watch. Before trying it, take a flower you can spare and examine it in good light. A hand lens is useful. The objective is to recognize the parts well enough that your fingers do not discover them by accident.

The green sepals sit outside the petals. Within the yellow petals, the anthers are arranged around the central female structures. The anthers produce pollen. The stigma is the receptive surface at the end of the style, which leads toward the ovary. You will need to leave the female structures undamaged while preventing the chosen recipient flower from supplying its own pollen.

Cultivated tomatoes commonly self-pollinate. That is convenient when maintaining a variety and inconvenient when trying to establish that a particular pollen donor supplied the cross. Putting two varieties next to each other is not a dependable way to make a known hybrid. Even if an insect transfers pollen, you have not thereby recorded which flower received it or which plant supplied it.

The timing is important. A flower that has already opened may already have shed pollen. The UC Davis Tomato Genetics Resource Center's crossing guidelines therefore use well-developed unopened recipient flowers. Their illustrated protocol is worth studying before handling a flower intended for an actual cross. We will return to the procedure after choosing the parents.

The fruit is not the next plant

Consider an ordinary cultivated tomato cross in which plant A supplies the recipient flower and plant B supplies the pollen. The flower and the fruit that develops around the seeds belong to A. The embryos produced by successful cross-fertilization carry genetic contributions from both parents.

Consequently, eating the fruit produced by your hand-pollinated flower is not a taste test of the future hybrid plant. You are sampling fruit tissue grown by the recipient. The new combination must be grown from seed before you can assess the fruit produced by that offspring.

This is easy to forget because the seed and fruit arrive in the same hand. A label on the fruit can therefore serve two purposes: it identifies where the fruit grew, and it identifies the intended parentage of the seed. Those are related records, but they are not interchangeable descriptions.

Use an explicit direction in your cross notation. Write the seed parent first, then the pollen parent: A × B. Record the individual plants as well as the variety names. “A plant 2 × B plant 1” tells you more than “A crossed with B,” especially if one parent is itself variable.

The reverse cross, B × A, should have a separate identifier. Even when two crosses seem similar in the traits you can see, their direction is part of the experimental history. Keep it rather than deciding in advance that it cannot matter.

One flower, one entry

Imagine that you prepare three flowers on A plant 2 and pollinate them with pollen from B plant 1. Call the attempts C01, C02, and C03. Some may fail to set fruit.

Each attempt needs a visible marker and a notebook entry. The entry should include the recipient, donor, preparation date, pollination date, and any problem you noticed. If the flower breaks, record that. If you later realize it was already open, record the uncertainty instead of silently treating the cross as confirmed.

Suppose C01 drops, C02 grows, and C03 loses its tag. C02 can remain in the planned project. C03 may still be edible and interesting, but its seed has an identification problem. You cannot repair the missing record by choosing whichever parentage you prefer.

Colored ties can help, but color alone is vulnerable. You may reuse it without remembering, or the tag may fade. A written cross number linked to a plant map gives you another way to recover the record. Tie labels loosely enough that growth is not constricted, and check them during normal garden visits.

Practice before the useful flower appears

Use spare flowers to learn how the anther structure comes away. Watch what happens when you pull too hard or grip in the wrong place. A practice flower can be damaged without costing the intended cross. Learn what an intact recipient should look like afterward.

Arrange the work where you can see it. Pollen from two donors should not share an unmarked surface. Tools and fingers that move between flowers can move material as well. A simple, orderly setup makes the parentage claim easier to defend later.

You do not need to finish all possible crosses in one session. Label what you have done before reaching for another flower. Check that the tag identifies both parents and the date.

The tomato you most want to improve is not automatically the best parent. It may combine the quality you love with several problems you have not yet measured. The other parent should bring a plausible contribution to the particular project, rather than merely looking different in a photograph.

Begin with plants you can grow and observe. If you cannot get both parents to flower during a usable overlap in your season, the planned cross has a scheduling problem before it has a genetic one. A small first project is easier when both parents are ordinary cultivated tomatoes that already grow reasonably well for you.

Write one sentence for the goal. Then write one sentence for each parent explaining why it belongs in the cross. If either explanation is mostly reputation, spend another season looking at the plant or choose a question you can answer with the evidence you have.

A worked parent comparison

Our six-space example seeks earlier ripe fruit without giving up an eating quality the gardener values. We will call the two candidates A and B. Here are the imagined observations that led to the choice.

The cross has an intelligible purpose. A contributes something the gardener likes to eat. B offers a reason to investigate earlier ripening. The table does not prove that the desired combination will occur, that earliness has been measured without environmental bias, or that either trait is controlled by one gene.

The entries should prompt questions. Were the transplant dates and plant sizes comparable? Was A planted in a cooler position? Did B's first ripe fruit count as a normal, usable tomato? Were the tastings repeated at similar ripeness? A tidy table can disguise an untidy comparison unless the notes travel with it.

Retain a copy of the starting descriptions. After a few years, the names A and B may become shorthand for what you hope they contributed rather than what you actually observed. The original record is a useful restraint on that gradual improvement of memory.

Choose constraints before choosing winners

Suppose an offspring ripens earlier than A but requires a support twice as tall as you can provide. Is that acceptable? If the answer is no, write it down now. Otherwise you may keep changing the rules to favor whichever plant has attracted your attention most recently.

Constraints should reflect the garden, not a catalog's ideal of excellence. “Must fit this support” gives you a decision you can make. So does a minimum usable harvest, provided you decide what enough means for your household.

Do not add so many constraints that you are effectively asking the cross to supply a finished commercial variety at once. Start with a small number of characteristics you can observe reliably. Keep the rest as notes. A note may become important later without being used to eliminate a plant this year.

Taste is a legitimate goal even though it is personal. You can improve the way you observe it without pretending to remove preference. Compare similar ripeness, use more than one fruit, and return on another day. If another member of the household dislikes your favorite, record the disagreement. The kitchen has supplied relevant evidence.

Health is not a badge you award after one summer

A plant that remains healthy while another suffers is worth noticing. It is not enough to establish resistance to a named disease. The plants may have faced different exposure, moisture, damage, or growing conditions. The diagnosis itself may be uncertain.

Even a familiar-looking blemish can belong to a disorder rather than an infection. Maryland Extension's account of blossom-end rot, for example, explains the role of calcium supply to developing fruit and environmental stress. A garden observation should not turn into a claim of inherited pathogen resistance merely because it involved damaged tomatoes.

Write what you saw before writing what you think caused it. “Dark sunken patch at blossom end on three early fruits” is more useful than “bad genes.” Photographs with a date and plant identifier can help a local adviser assess a problem. You can continue taking notes while withholding a diagnosis.

Do not deliberately bring diseased material into a home garden to challenge a candidate. This book's selection examples assume normal garden observation, not a pathogen-screening program. Formal resistance testing requires appropriate methods and containment, and its results should not be imitated by inference.

For a first cross, choose parents you can identify, keep healthy, and describe without borrowing every adjective from their sellers. Familiar parents give you a clearer basis for judging their offspring.

Set out the labels before the tools. The moment after a successful-looking pollination is a poor time to discover that the marker has dried up.

Choose a manageable number of attempts and give each one an identifier. Put the intended recipient and donor in the notebook before touching the flower. Check the plant labels against the map. This sounds fussy until two pots are moved to make watering easier and the arrangement you thought you would remember no longer exists.

Use a work surface that can be cleaned and a setup that keeps donors separate. You need a way to see the flower, fine tools suited to the task, and tags that will remain readable outdoors. If a particular tool makes you crush the flower to use it, practice with something else.

The short procedure

The botanical procedure below is a concise account of the UC Davis TGRC method; use its photographs when learning the manipulation.

1. Select a well-developed recipient bud that has not opened. A bud showing some yellow can be receptive that day; a younger one may need another day or two.
2. Expose and remove the petals and anther structure carefully, leaving the central female structures intact. Avoid spilling the recipient's pollen onto the stigma.
3. Obtain visible pollen from a mature flower of the intended donor and apply it to the recipient stigma.
4. Mark the flower immediately. Record the parents and dates. Clean hands and tools between crosses.
5. Revisit the flower. Fruit development suggests that fertilization occurred, but does not by itself establish the donor's identity.

A swelling fruit cannot tell you whether you prevented self-pollination successfully. Parentage depends on the procedure and the record, with later observations sometimes providing additional evidence.

Keep one donor in play

If this is your first attempt, simplify the table in front of you. Work with one donor at a time. Finish its labels and notes before bringing out another donor's flowers. A sophisticated coding scheme cannot recover information lost in a mixture of pollen.

Our example uses the cross number C02. The tag on the flower says C02; the notebook says “A2 × B1,” with separate preparation and pollination dates. The map identifies A2 and B1. If the tag is small, it need not contain the whole history. It does need a dependable link to that history.

Record a second pollen application as a second event on the same attempt, not a new cross with the same number. If you accidentally use a different donor, do not quietly overwrite the earlier entry. Mark the attempt uncertain and keep it out of any group you later describe as having confirmed intended parentage.

Keep what happened distinct from what you intended.

When the flower falls

Flowers sometimes drop after a crossing attempt. The flower may have been damaged. The pollen or recipient may not have been in a suitable state. Weather and plant condition also matter. Extension discussions of vegetable responses to heat provide useful context for poor fruit set; a hot spell is not proof that your hand technique alone was responsible.

Write the outcome and any obvious circumstances. If all attempts made during one session fail, compare the common features before declaring the two parents incompatible. Did you use the same stage of recipient bud? Was the pollen visibly available? Did the tool repeatedly damage the central structure? Was the plant already struggling?

Change one practical element at a time when possible. Trying a different recipient stage, different tool, different donor, and different handling method together may produce fruit, but it will not tell you which change helped. A small garden does not need laboratory precision to benefit from an orderly comparison.

A few attempts will teach you more about handling flowers than about the compatibility of two varieties. Try again when the plants are growing well and the weather is suitable.

Protect the record until harvest

Check tags while you water. A marker may loosen, become hidden, or press against a growing part. A photograph that includes the fruit, its tag, and the plant can serve as a backup, but it should supplement the written record rather than replace it.

When a tagged fruit is ready, carry it in its own labeled container. Do not put it into the general harvest bowl and plan to retrieve it later by shape. The seed will soon look much like seed from the other tomatoes. You have reached the point where keeping things separate becomes more important than recognizing them by sight.

Keep the fruit’s cross label attached through harvest. Mixing two crosses in one unmarked bowl loses the parentage record.

A successful cross ends its first season as a packet, not a tomato. The packet is small enough to misplace and durable enough to outlast your memory of the summer. Treat its label as part of the material.

Before opening the fruit, copy the cross identifier onto the processing container. Put a second label in the notebook or on the drying station. Work on one identified lot at a time if the kitchen is busy or the available surface is small. A tray full of matching unmarked jars is an avoidable problem.

Harvest seed from ripe fruit belonging to healthy, identified plants. If there is a suspected seed-borne disease, do not assume that routine washing or fermentation makes the material suitable to share. Minnesota Extension's bacterial-canker guidance explains why infected seed and transplants can matter even when contamination is not obvious. Ask for an appropriate diagnosis rather than treating a seed-cleaning step as a universal cure.

Separate the gel, then dry the seed

Seed Savers Exchange describes several extraction methods in its tomato guide. For a small labeled lot, remove the seeds and surrounding pulp into a container, allow a short monitored fermentation to loosen the coating, then separate and rinse the seeds through a suitably fine strainer. Spread them in a single layer with air circulation until thoroughly dry. The guide illustrates the stages; the rate depends on conditions, so a neglected jar is not improved by extra days of waiting.

Keep this operation separate from food preparation. The fermenting pulp is being used to process seed, not to make something to eat. Clean the surfaces and tools afterward. In a household with children or animals, place the containers where they will not be mistaken for a snack or knocked into one another.

Do not seal damp seed into its final storage container. Nor should a home gardener use excessive heat to hurry the job. Iowa State's seed-storage guidance specifically cautions against food dehydrators that can become too warm for seed. Give the drying station time and keep the identifying label with it.

Keep lots separate for a reason

A seed lot is a group you handle under one recorded identity. At the beginning of a breeding project, a useful lot may be the seed from one tagged fruit. Later, you may choose to combine seed from several fruits or plants. Combining is a decision that should appear in the record.

Suppose C02 and C05 are separate fruits from the same intended parental combination. You could keep them apart and note that they share the same parents. If one later shows an unexpected result, the separate lots leave you something to investigate. Mixing them immediately saves an envelope but removes that distinction.

Do not confuse a fruit identifier with a generation label. “C02 fruit” describes the harvested object. “C02 F1 seed” describes the intended generation that will grow from it. After an F1 plant produces selfed seed, that new seed belongs to F2. The packet should make clear whether a number describes the plant that bore the seed or the generation you will sow.

We will use the generation-to-be-grown on packet fronts. For example: “C02 — F1 — A2 × B1 — harvested 2027.” Put more detailed handling notes in the notebook, linked by C02. Keep the harvest year on the packet even if you sow in a different calendar year.

A label you can read next winter

At minimum, preserve the project or variety name, lot identifier, source plant, generation where relevant, harvest date, and any uncertainty about pollination or handling. Write with something suited to the storage conditions. Check that the packet is closed and that small seed cannot slip through an unsealed fold.

Take a photograph of the packet and its matching notebook entry. This is an inexpensive backup to the information. It is not a backup to the seed. Later, if the project becomes worth preserving, a second properly identified packet stored separately can protect the material as well.

If you discover a mistake, write the correction in a way that leaves the original uncertainty understandable. “Label found detached; possible C02 or C05” is less satisfying than a confident identifier, but it protects the rest of the project from false certainty. You may still grow the seed as an unconfirmed experiment. Keep that experiment separate from the line whose history you claim to know.

There is no need to save every seed a plant produces. Estimate the next sowing, a germination test, and a reasonable reserve, then keep a quantity you can manage. Saving an enormous unidentified bulk does less for the project than a small, dry, well-labeled packet.

Put the packet away only after checking it against the notebook. By then, the fruit has been eaten or discarded and the plant may be nearing the end of its season. The next year's work will begin with whatever this envelope can tell you.

The first seedling unfolds two seed leaves and may look much like any other tomato seedling. Its cross label supplies information that its appearance cannot yet provide.

F1 means the first filial generation from the cross being discussed. It is a place in a pedigree, not a quality rating. If the cross succeeded as intended, these plants are the first offspring of your recorded parents. Whether they are useful is a separate question.

Grow more than one when your space allows, and keep individual plant numbers. A packet can give several plants the same parental origin without making their positions, treatment, or eventual records interchangeable. A plant that loses its label should not inherit the nearest convenient number.

Keep the parents in the comparison

If you want to know whether an offspring improved on A, grow A alongside it when practical. A memory of last year's tomato is a weak control. The weather, planting date, and your own preferences may have shifted. A parent grown in the current season gives you a more relevant reference.

In our six-space plan, the second year might include one plant of each parent, two F1 offspring, and two familiar kitchen varieties. That is still a small demonstration, not a breeding population large enough to explore every possibility. Its purpose is to learn whether the cross is worth taking forward and to obtain seed for the next generation.

Similar does not mean proved

When two sufficiently uniform, genetically fixed parent lines are crossed, the resulting F1 can be relatively uniform in the genetic features that distinguish those lines. Garden material is not always fixed in that way. A parent saved from a variable population, or from a commercial hybrid, can give a less uniform set of offspring.

Do not use uniformity alone as proof that every seedling came from the intended cross. Selfed offspring of a uniform recipient may also resemble one another. Conversely, a difference between two plants can result from growing conditions rather than mistaken parentage.

Sometimes a visible parental trait helps you check a cross, but it must be interpreted with a known inheritance pattern. Choosing a marker simply because one parent looks different is not enough. A trait may be hidden in the F1, influenced by several genes, or changed by the environment. The notebook should distinguish “consistent with the intended cross” from “parentage established by a suitable test.”

For a first home project, the best available evidence may be a careful crossing procedure, dependable labels, and observations that fit the intended history. That can be enough to continue an experiment. It is not the same as independent genetic verification, and it should not be described as such when the seed is shared.

Observe the plant, not just the best fruit

Return to the goal you wrote before making the cross. Record first flowering, first ripe usable fruit, harvest observations, plant habit, and repeated eating impressions at a level you can sustain. A notebook that demands an hour every evening will probably go blank during the busiest part of the summer.

A few clearly defined observations beat a large number of impressions that change meaning from week to week. Decide what counts as ripe. Decide whether yield means every fruit picked or only usable fruit. If you change the definition, mark the change rather than making the early and late entries appear directly comparable.

Photograph the plant as well as the fruit. A close-up of a tomato cannot show the support it needed, the foliage left at harvest, or the amount of space it occupied. Those features may decide whether you want to grow its descendants, even when they do not make an attractive catalog image.

Keep a separate entry for unusual treatment. If a stem breaks and you retrain the plant, note it. If watering fails for several days, note that too. An offspring does not need a spotless season to teach you something; it needs a record that allows its results to be interpreted.

Save the next generation deliberately

To produce F2 from an F1 plant in this project, save seed from self-pollinated flowers on that F1. Use an appropriate isolation method when maintaining control of the pedigree. Seed Savers Exchange describes blossom bags placed before flowers open as one method of excluding insect-carried pollen. Mark the protected fruits so that the later harvest remains identifiable.

Keep seed from different F1 plants separately at first. Write the plant number and the generation to be grown on each lot. This preserves your ability to compare families if something unexpected appears next season.

An excellent F1 may be enjoyable in its own right. Its selfed descendants do not necessarily reproduce it. If you want the same F1 combination again, that generally means remaking the original cross from the appropriate maintained parents. Selecting descendants toward a stable line is a different route, and its result need not be an exact copy of the F1 you liked.

The packet going into storage now has a new question attached: which combinations will appear when this first-generation plant produces offspring of its own?

The F2 is often the part of a small breeding project that earns the most attention. Plants from the same recorded cross can differ in visible ways. The differences are useful, but the number you can grow limits what you will see.

Before looking at real traits, use a deliberately simple model. It will let us calculate something exactly and then identify the assumptions that make the calculation work. The model is not a claim that tomato flavor or earliness follows a single-gene pattern.

Two copies at one position

Suppose a genetic position has two versions, written A and a. An individual can have AA, Aa, or aa at that position. These symbols describe versions and combinations; the capital letter is not an award for being better.

Cross an AA parent with an aa parent under the simple model. Each offspring receives A from one and a from the other, giving Aa. When an Aa individual self-pollinates, each reproductive contribution can carry A or a. The four equally likely pairings can be written in a small table.

The expected proportions are one quarter AA, one half Aa, and one quarter aa. If A is completely dominant for the particular observable trait, AA and Aa have the same phenotype for that trait. Genotype describes the genetic combination; phenotype describes what is expressed or observed. OpenStax's discussion of inheritance explains this framework and the limits introduced by more complicated patterns.

These are probabilities for offspring under stated assumptions. Four seedlings do not have to distribute themselves politely into one, two, and one. A small sample can miss a class altogether. Nothing in the arithmetic allocates a recessive plant to the fourth pot because the first three have used the other possibilities.

What twelve spaces can miss

In the same model, the chance that one offspring is not aa is three quarters. If offspring are independent with respect to this event, the chance that none of n offspring is aa is (3/4)^n. The chance of seeing at least one is therefore 1 − (3/4)^n.

Here are calculations for that toy model, rounded to the nearest whole percent. They are not recommended population sizes for a real breeding objective.

Now ask for two particular recessive combinations at two independently inherited positions. If the chance of each is one quarter, the joint chance is one sixteenth. With twelve offspring, the chance of seeing at least one individual with both is only about 54 percent. With twenty, it is about 72 percent. A modest extra requirement can consume a surprising amount of garden space.

The calculation becomes less convenient when the positions are linked, the traits involve many genes, viability differs, or observations do not cleanly reveal the genotype. Those are ordinary biological possibilities, not reasons to abandon the project. They are reasons to avoid treating a few seedlings as a complete inventory of what the cross could produce.

A worksheet, not a prophecy

You can use the formula to understand sampling. You cannot insert a guessed probability for “excellent taste” and turn the result into a trustworthy promise. To use a probability meaningfully, you need a defensible model or data relevant to the event you are counting.

A practical implication survives without an exact probability: limiting the number of offspring limits the combinations you have a chance to inspect. If you can only grow six, choose a modest question and keep seed in reserve. You may decide to sample the same generation again in a later season, rather than pretending the first six exhausted its possibilities.

Do not mix up more seed with more observed plants. A thousand seeds in a packet represent potential samples. They do not provide the information you would have obtained by growing and evaluating a thousand plants.

Why a good-looking plant can still vary next year

Return to the single-position model. An AA plant and an Aa plant can look the same for a completely dominant trait. After selfing, AA remains fixed at that position under the model, while Aa can produce several combinations. Appearance alone has not told you which plant you selected.

Growing descendants separately allows you to observe what each selected parent produces. That is one reason to keep families identifiable. If you mix all selected seed into one envelope, variation in the next generation becomes harder to connect to a particular parent.

Selfing tends to reduce heterozygosity across generations in a simple model, but a generation label is not a certificate of complete uniformity. An F6 family may still vary in a trait you care about. Selection, accidental crossing, the starting material, and the parts of the genome involved all affect the result.

The model explains why descendants need testing. It does not make flavor a simple trait or oblige four seedlings to display a tidy ratio.

Keep the question small enough to answer

For the first variable generation, choose a limited set of observations tied to your original goal. Number every plant before the interesting differences become apparent. Record the plants you discard as well as those you keep, with a brief reason. Otherwise the record will eventually contain only success stories and lose the information about what the cross tended to produce.

You may find a plant you like for an entirely different purpose. Keep it as a separate branch if you have room. Give that branch its own goal and identifier. Record why you changed the goal and retain the earlier notes.

By midsummer, the project can become a negotiation with a few plants. One ripens early but tastes ordinary. Another is delicious and awkward to support. A third has done nothing conspicuous except keep producing fruit you use.

The last plant deserves a proper entry. Novelty attracts attention, and a breeding notebook should correct for that bias rather than amplify it.

Start with the same observations for every candidate. You can take additional notes about a striking plant, but do not give only that plant a full inspection. A comparison becomes lopsided when the favorite is measured and the rest are remembered.

Define an observation once

“Early” needs a definition. For this project it might mean the date of the first ripe, usable fruit after a shared transplant date. That definition will not answer every question about earliness, but it tells you what the number in the column means.

“Yield” needs a boundary too. Are you counting fruit, weighing fruit, or recording a rough household-use category? Are split or otherwise unusable fruits included? A plant with many tiny fruits cannot be compared fairly with a large-fruited plant by count alone if your real concern is usable food.

You do not have to measure everything in grams. You do need to avoid changing the unit midway through the argument. A rough, consistently used measure can support a rough decision. It should not be dressed up as a precise agronomic result.

Our example gardener keeps five columns: first ripe date, repeated taste notes, usable harvest, plant fit, and problems. None produces a final score automatically. The main goal is earlier fruit with acceptable eating quality; the other columns help identify costs that the headline improvement might conceal.

Taste without letting the label do all the work

Taste similar samples at similar ripeness. If possible, have someone code the pieces so that the name is not visible during the first comparison. Use more than one fruit and repeat on another day. You are trying to reduce avoidable influences, not to turn a home kitchen into a sensory laboratory.

Write a short description before awarding a number. “Pleasant acidity, firm skin, would happily eat again” carries information that a score of four does not. A numerical score may help sort a large sheet later, but only if you know what the numbers meant when you used them.

Avoid averaging away a real household disagreement. If one person wants a sharp tomato and another wants a mild one, there may be two useful outcomes. The aim is not to discover which person has correctly calibrated taste. It is to choose fruit for the people who will eat it.

Tasting a single exceptional fruit can give a plant an undeserved reputation. Mark the occasion and return. Conversely, one disappointing sample is not necessarily enough to eliminate an otherwise useful candidate. The time you can devote to repeated observation is one of the resources limiting the project.

An example decision

Here is the example gardener’s decision sheet.

The fourth entry matters. A compromised comparison is not the same as an unfavorable genetic result. You may still lack room to keep that plant's descendants, but the reason for the decision should remain clear.

Write a sentence explaining every advancement. “Keep because it ripened earlier while remaining acceptable in three tastings” is more useful than “winner.” A future season can test the sentence. The word winner mostly records your mood at harvest.

Keep families rather than an anonymous trophy packet

Save each chosen plant's seed under its own identifier. When descendants grow next year, you can ask whether the useful combination persists within that family. If it does not, the project has still supplied information about the selection you made.

The amount of uniformity you want depends on the aim. A line intended to reproduce a fairly specific type needs repeated attention to consistency. A deliberately variable garden population has a different purpose. Both can be worthwhile, but they should not be described as the same product when shared.

Do not eliminate variation merely because it makes the row less tidy. Eliminate or retain it according to the goal. A difference in fruit shape that does not bother you may be irrelevant; a difference in ripening time may be central. The notebook should show which distinctions you decided to care about.

Finally, keep some seed from the earlier generation when possible. A reserve gives you a way to revisit a decision. It cannot undo the lost season, but it can keep one narrow selection from becoming the only route left to the project.

The warmest corner of the garden can make a plant seem unusually early or productive. So can a better root run, a less damaged transplant, or the fact that you remember to water the container nearest the tap.

You cannot remove every difference from a home garden. You can avoid arranging the trial so that one family receives all the advantages. Sketch the beds, note shade and wind, and place comparable plants so that their positions do not perfectly match their identities.

If all descendants of your favorite occupy the good end of a bed and all descendants of its rival occupy the poor end, the result will be difficult to interpret. Moving labels around in a spreadsheet afterward cannot separate the family effect from the position effect. The opportunity to improve the comparison was before planting.

Repeat a small comparison

Suppose you have room for two plants from family X and two from family Y. Arrange them across the available positions rather than placing both X plants together in the best spot. This is a modest attempt to reduce obvious bias, not a claim that four plants provide a definitive trial.

If the site has two reasonably comparable pairs of positions, assign one X and one Y within each pair. Use a coin or another simple random choice to decide which goes where, and record the arrangement. Random assignment does not make the positions identical; it prevents your preference from always deciding who gets which one.

A real garden may force exceptions. A large plant may need a support that only fits in one location. A container may be too heavy to move. Write the constraint down. You can still observe the plants, but you should be cautious about attributing their differences entirely to heredity.

Repetition across seasons helps because the growing conditions will change. It does not mean that a family performing well twice has become universally adapted. It means you have observed it under two sets of circumstances rather than one.

Do not count the same evidence twice

Ten fruits from one plant are ten fruit observations, not ten independent plants. They can tell you something about that plant's fruit, including its consistency. They cannot replace ten plants when the question is how a family behaves across individuals.

Likewise, five photographs of the same diseased leaf do not constitute five separate disease events. Decide what your unit of observation is before collecting a large set of numbers. A spreadsheet becomes misleading quickly when its rows look independent but describe repeated views of the same thing.

This matters when you summarize. “All ten tasted samples were acceptable” and “all ten plants produced acceptable fruit” are different claims. Keep the distinction in the prose you write for a future seed recipient. Precision need not make the description long.

Weather belongs in the record

Note events that plausibly affected the comparison: a heat spell, hail, prolonged rain, a missed irrigation, or an unusually early cold period. You do not need a private weather station to write “watering line blocked for three days.” That note may explain more than an elaborate graph of a distant station's temperature.

Temperature can affect flowering, pollination, and fruit development. Maryland Extension's discussion of pollination in a changing climate is a useful reminder that a plant's failure to set fruit is not automatically evidence of a failed cross or poor inherited yield.

Avoid correcting every observation with an explanation that protects the favorite. If you excuse X because the season was difficult, consider whether Y faced the same difficulty. Record the event first and interpret it consistently.

A plant that performs reasonably under an awkward season may be valuable to you. The restraint is in how you describe it: “It gave usable fruit during this hot season at this site” is defensible if that is what you observed. “Heat-proof” is a much larger claim.

Make the next decision, not the final pronouncement

At the end of the comparison, ask what the evidence is good enough to decide. It may justify keeping a family for another year. It may justify giving a small amount of seed to a collaborator with a clear description of its experimental status. It may not justify announcing a stable, widely adapted new variety.

Use a short review sheet. What was the original question? Which observations bear on it? What spoiled the comparison? What would be worth repeating? What can you stop doing because it did not help?

The last question saves work. A measurement that was tedious and never influenced a decision may not deserve another season. Conversely, the one date you forgot to record may become the first entry on next year's sheet.

Keep the comparison small enough to finish. Two families with complete notes may tell you more than a larger planting you cannot keep track of.

Seed stored badly can make a careful breeding history unusable. The envelope may still carry every parent and date, but the living material inside it has its own requirements.

Keep cleaned seed dry, cool, and protected from pests. Label individual packets and place them in a suitable closed storage container. If you use desiccant, keep it separate from the seed and food. Check the storage method against a horticultural source rather than improvising with a hot appliance or an untested drying shortcut.

The Iowa State seed-storage guide also describes a germination test: sample seeds, place them on moist paper in a suitable warm environment, keep them identified, and count those that germinate. The temperature and observation period must suit the crop. The number germinated divided by the number tested gives the observed germination proportion.

A test is useful because appearance and age alone cannot tell you exactly what a packet will do. Its result is still an estimate from a sample, not a guarantee for every remaining seed or for emergence in soil.

Work with the number you actually tested

Suppose 16 of 20 seeds germinate. The observed result is 80 percent. It is not proof that exactly 80 of the next 100 will germinate, and it does not include the losses that can occur after germination.

If you need ten seedlings and use 80 percent as a rough planning estimate, dividing ten by 0.8 gives 12.5 seeds. You must sow a whole number, so thirteen would only be the rounded estimate, not an insurance policy. A sensible plan may include a few more if the seed supply and space permit. The test's uncertainty and later losses still matter.

A very small test has limited resolution. Nine out of ten and eight out of ten differ by one seed, even though the percentages look ten points apart. Do not turn that small difference into an elaborate story about one storage method being superior without a better comparison.

For a rare packet, testing consumes part of the reserve. Decide how much information you need and how much material you can spare. A method appropriate to abundant seed may be wasteful when only a few seeds remain. Seek specialist advice for irreplaceable material rather than treating a generic home test as the only possible choice.

Keep an inventory that answers ordinary questions

An inventory should tell you what the packet is, where it is, and what you can reasonably do with it. Start with a lot identifier, generation, harvest year, approximate quantity, storage location, and the most recent germination observation if there is one.

A count does not have to be exact when you have plenty. “About one hundred” may be sufficient for planning a twelve-plant sowing. When the packet holds seven seeds, count them. Precision should follow the decision, not the desire to fill every cell in a sheet with a number.

Link the lot to the plant record. The inventory is not the place to repeat every taste note or weather event. It should lead you to that information. A compact system is easier to update when seed is removed, divided, or given away.

Record transfers as they happen. If you send half a packet to someone else, note the amount, recipient, date, and the exact identity used. If that person later reports an unexpected result, you will know which material they actually received.

Back up the seed and the information separately

Photographs and files protect the record. A duplicate packet protects the material. Neither substitutes for the other.

If a line becomes important, consider keeping a small reserve in another suitable location. The value of separation is that one leak, pest problem, misplaced box, or accidental disposal need not remove everything. A duplicate stored in the same vulnerable drawer has a more limited role.

Do not distribute seed merely to create a backup without discussing what the recipient is agreeing to do. Keeping a packet, growing a trial, and maintaining a line are different commitments. Ask for the one you need in plain language.

Review the drawer before buying more seed. You may already have the next useful experiment. You may also discover that a project no longer interests you. That is a legitimate reason to stop maintaining it, provided you do not silently pass it off as something more established than it is.

Leave an intelligible final note for a discontinued project: what it was, why it stopped, and whether any material remains. That note distinguishes a deliberate ending from a missing envelope.

A plant worth keeping is the beginning of another observation, not the end of the breeding process. Its descendants show what it passes on and how consistently the qualities you selected reappear.

Keep families separate while those questions remain important. Grow more than one descendant when you can. Observe the traits you selected, retain useful material, and record departures from the intended type. A promising plant that produces a wide assortment next year has not failed to obey its label; the label simply did not yet describe a stable result.

Uniformity has a purpose rather than an absolute moral value. A gardener may deliberately maintain a variable population for a range of fruit types. Someone receiving a named line with a specific description may expect more consistency. The problem is making one kind of material sound like the other.

Generations take space and time

In an outdoor project with one generation per year, the first crossing season produces F1 seed. The next season grows the F1 plants. The following season grows F2. Each later selfed generation requires another cycle of plants, observations, and seed.

This schedule is an example for a single annual growing cycle. Other environments and facilities can allow different schedules. More cycles alone do not replace careful evaluation in the conditions where you want the finished plants to grow.

There is no generation number at which a notebook can automatically print “finished.” Some traits may be consistent while others remain variable. State what you have observed and over how many generations rather than relying on a large F number to carry the whole claim.

Keep earlier seed where practical. If a later selection loses something important, the reserve may let you revisit another branch. The reserve also preserves material that has not passed through every choice you made along the way.

Maintaining a variety is a separate population question

One tomato plant can produce viable seed, but that fact does not answer how best to maintain a variety over time. Seed Savers Exchange recommends saving from five to ten plants for ongoing variety maintenance. That recommendation is distinct from the number of F2 plants needed to sample a particular breeding objective, which depends on the traits and probabilities involved.

Our six-space worked garden is therefore a deliberately limited learning project. It cannot simultaneously serve as a large variable-generation trial, a broad preservation population for several varieties, and a complete summer food supply. When your goals grow, the space plan must change too.

Collaboration can help, but it also changes the record-keeping problem. Agree on plant identities, observations, and how seed will be kept separate before distributing packets. A collection of enthusiastic anecdotes is pleasant to receive; it is harder to compare than a small set of shared observations.

If collaborators use different climates and methods, keep that context. Differences between their results may be important. Do not average them into a single score that conceals the conditions under which a family did well or poorly.

Write a seed description you can defend

A useful description includes the recorded parentage, generation, traits observed, main limitations, and whether the material remains variable. Give the recipient enough information to decide whether the experiment suits their garden.

For example: “C02-F4 family from the recorded A2 × B1 cross. Selected for earlier ripe fruit with acceptable eating quality at one site. Some variation in fruit size remains. Observed through four growing seasons; no formal disease-resistance testing.” The identifiers and history here are illustrative.

Do not call seed disease-free merely because the source plant looked healthy. Do not claim a precise parentage for a lot whose tag was lost. Do not attach the name of a preserved variety to deliberate hybrid descendants without making the cross clear. These are practical ways to keep a recipient's next season from beginning with the wrong premise.

You can share a useful experiment without announcing a finished variety. In fact, a clearly described unfinished family may be exactly what another gardener wants. The record gives them a place to begin, and gives you a way to understand what they send back.

The worksheets below are deliberately short. Use them on paper, in a spreadsheet, or in another system you already maintain. A pencil is enough.

Download the blank garden workbook for five CSV sheets. They open in ordinary spreadsheet software. The printable notebook has the same records laid out for handwriting. Both are blank; the example identifiers in this book are not observations from your garden.

Before the season

Write the main goal and the available space first. Then list the varieties or families you intend to grow, the role of each, and the limits you already know. Include the ordinary kitchen plants. A breeding plan that omits the household's harvest needs may be abandoned for understandable reasons when planting time arrives.

Draw a map. Give each plant position an identifier independent of the variety name. This lets you record a move or replacement without pretending the physical position and the genetic material were always the same thing.

Decide which observations you will collect. Define first ripe fruit, usable harvest, and any taste scale in ordinary words. You can revise the plan later, but date the revision.

At the flower

Before preparing a recipient, make the cross entry. Confirm the seed parent and pollen parent, including individual plant numbers. Give the attempt a unique identifier and prepare its tag.

Record preparation and pollination as separate events if they occur at different times. Add a second pollen application to the same record. Mark uncertainty, damage, or an accidental change of donor while you still remember it.

When the flower sets fruit or drops, record the outcome. The unsuccessful attempts belong in the notebook too. They may help you improve technique and prevent an exaggerated impression of how reliably the method worked in your hands.

At harvest

Keep selected or crossed fruit separate and identified. Transfer the identity to the processing container before opening the fruit. Transfer it again to the drying station and final packet. Each move is an opportunity for the label to become detached from the material.

Give each seed lot a clear generation-to-be-grown label. Connect it to the source plant and cross history. Record any combining of lots rather than making the new bulk appear to have always been one sample.

At the end of the season

Review the goal against the observations. Choose which families to advance, which to keep in reserve, and which to stop. Write a reason that can be understood next year.

Check the seed inventory against the physical packets. Note approximate quantities, storage locations, and any germination tests. Keep a copy of the records outside the place where the only paper notebook lives.

Finally, make next year's space budget before making more crosses. The most attractive page in a breeding notebook is often the one headed “Next year.” Give it the same arithmetic you gave this year's garden.

A record should be easy to continue when you are tired and the garden needs watering. Leave room for a sentence. A number can tell you that something changed; the sentence can tell you that the support fell over the night before.

The next season does not have to make the project larger. You might keep one family, grow a dependable variety beside it, and put the rest of the space back into tomatoes you already enjoy. That is a usable result from a small garden.