Saturday, February 13, 2016


We recently collected a limited supply of semen on two “outcross” pedigree Jersey sires who have polled heads and possess the most desired Casein (protein) genes.

The goal in this outcross selection was to produce a Jersey with superior constitutional strength, more resilient udders, strong natural fertility, and ultimately more longevity of production.     Our current herd has enough longevity that we only need to replace 10% of our cows annually.

Keep in mind, our herd is managed as a grass dairy—heifers raised on grass, cows fed a primarily grass ration (rotation grazing in season, wrapped bale long stem winter forage) limited to 6 to 8 pounds of grain daily—imitating a typical organic dairy design.

For developing organic milk markets, the desired Casein gene is “A2 Beta Casein”.   We have identified several such sires from outcross bloodlines, and are linebreeding them.  
Nutrition specialists believe A2 Beta Casein (found active in fluid milk, yogurt, cultured cream) offers many health benefits, especially to those suffering autoimmune diseases.

For dairies pursuing cheese making the desired Casein gene is “B Kappa Casein” (cheese yield increases up to 15%).    For dairies pursuing yogurt making the desired Casein gene is “A Kappa Casein” (flavor enhancement).    

For dairies tired of dehorning, naturally polled offers an escape from a dirty job that has been proven to set back calf growth rates.    We have some animals that are homozygous (“pure”) polled.     

How much is semen worth?

From the Oct-Nov 2013 Conceptions Dairy Route Leter

The platitudes are plentiful—“the bull is half your herd” for example—and the majority of population geneticists basically see cows as unknown ciphers, expecting all herd improvement to come from sires selected for herd mating.     Thus they promote “composite ranking indexes” to identify the “best” bull to use, no matter the cows being mated.    Of course, the higher the ranking, the higher the price.

Herd improvement is half “selection” and half “mating”.

In spite of the population genetics view, your cows are not ciphers—they are individual bundles of genes as well as accumulations of mitochondrial capability.     If the bull is half the herd, your cows are clearly the other half, and have equal impact on the results from sire selection.

Thus the development of “mating programs”, including the aAa “breeding guide”.      While evaluation data on AI sires indicates their average performance to date against contemporaries, it cannot answer the question—Will I get an average, a better, or a worse daughter of this bull?     The aAa Breeding Guide is designed to address that question, knowing you can only breed a cow to one bull at a time.   Average results across the population (such as the theory of group success from Genomics) do not determine our specific results from how we choose to breed old Bessie or her young daughter Bertha.

An example of the use of “aAa” concepts without analyzing your cows

The dominant sire influence in today’s commercial Holstein herd is O-Bee Manfred Justice, popularly known as O Man.       If he had arrived ten years earlier, before the use of health and fertility traits for sire rankings, he would likely have been culled on his minus type data, especially for udder traits.   But he arrived after the Net Merit index was recalculated to emphasize “fitness” traits and that saved him.

“O Man” under the “aAa” system is a source of Strong (4) front ends, Open (3) skeletal structure and Smooth (5) udder and body conformation.     On cows who needed this, he produced long lasting cows who milked well, bred back easy, and last long enough to produce replacements.    On cows who don’t need this, he produced cows with deep, meaty udders and short legs, not as milky and easily culled.

If you have a lot of “O Man” influence you start looking for bulls who can add Dairy (1) to front end and rib structure, Tall (2) for softer udder texture, higher rear udders and more stature, and Style (6) to blend the skeletal structures together and keep the hind legs squarely under the body.     Checking out Holstein USA “Red Book Plus” sire data, it seems there are very few such bulls currently active in all the commercial AI systems!    (O-Man is an even bigger influence on AI sires than on our cow herds.)

We offer four of the twelve active evaluated AI Holstein sires offering 1-2-6 aAa characteristics.

One has been in our lineup a long time, and many of you already milk his daughters.   The other three are more recent, and I want to bring them to your attention, even though they cost a bit more    Thus I have a comparison chart on the following page to explain my thinking.

Note that of the four,  6HO1159 Regancrest Mac Bence  is in fact the #1 sire for TPI at 1945, while the #2 sire for TPI at 1854 is  203HO 376 Probstland Ernesto.    Meanwhile, the #1 Type sire in this aAa group is the newest entry,  566HO1192 Mr Apple Jack Red.     So on various rankings, these three are a pretty elite group  (not as obvious when all you look at are the unsorted ranking lists). 

Comparison:   four sires, with same “aAa” qualities, but different trait patterns

Who is the best buy for your herd?

Overall Type           Udders                       Feet & Legs               Milk  volume           Holstein TPI

Apple Jack   +2.97    Bence           +2.74     Apple Jack   +2.29     Bence          +698      Bence          1945
Bence           +2.31    Apple Jack   +2.56     Bence           +1.95     Ernesto        +446     Ernesto        1854
Ernesto        +2.14     Ernesto        +2.19     Ernesto         +1.53     Popular        -215      Apple Jack  1599
Popular        +1.06     Popular        +0.96     Popular        +0.63     Apple Jack   -675      Popular        1463

Productive Life   Dtr Preg Rate     Somatic Cell        Calving ease         Mat Stillbirths     $NM rank

Popular       +3.3   Popular      +1.6    Bence        2.79    Ernesto     5.6%     Popular   7.1%    Bence     $386
Bence         +2.4   Bence         +1.1    Popular     2.85     Popular    7.0%     Bence      7.4%    Ernesto   $314
Ernesto       +0.6   Ernesto       -0.3    Ernesto      2.89    Bence       8.8%     Ernesto    8.5%    Popular   $ 78
Apple Jack  -1.1   Apple Jack  -2.0   Apple Jack 2.95    App Jack 10.5%    App Jack 9.1%    App Jack $ 12

%butterfat               Bfat volume        %protein              Protein volume      Lethal Recessives
Apple Jack  0.20%    Bence        +30    Ernesto    0.08%     Ernesto      +33       Bence              (none)
Bence          0.02%    App Jack   +26    AppJack  0.08%     Bence        +15       Ernesto            (none)
Ernesto        0.02%    Ernesto     +22     Bence    -0.02%     App Jack   +  0       Popular           HH2C* 
Popular      -0.04%     Popular     -18     Popular  -0.04%     Popular     - 17       Apple Jack       CVM* 

Summary of how these four rank against each other:

Bence        #1  (6)   #2  (7)   #3   (2)   #4   (0)          My overall #1 pick for breeding cows needing 1-2-6
Ernesto     #1  (3)   #2  (3)   #3   (9)   #4   (0)          My overall #1 pick for heifer calving ease and 1-2-6
App Jack  #1  (3)   #2  (3)   #3   (2)   #4   (7)          A specialty type bull for Red & White show herds
Popular     #1  (3)  #2  (2)   #3   (2)   #4   (8)           Not as good as the other three for too many traits

Comparing them on price:   I am assuming five straws gives you a live heifer calf on the average
Bence      (List $28)  our price $14             Investment per heifer calf:  $70   add raising costs   $  1530
Ernesto   (List $20)  our price $12             Investment per heifer calf:  $60   add raising costs   $  1520
App Jack(List $30)  our price $18             Investment per heifer calf:  $90   add raising costs   $  1550
Popular   (List $14)  our price $ 9             Investment per heifer calf:  $45    add raising costs   $  1505


The dairymen who are losing the most money on their breeding programs are those who treat semen as if it was a commodity and demand the lowest price possible.    AI studs respond by selling their lowest total value sires.    In our program, the difference between our best sires and our cheapest sires may only add $25 per fresh heifer to your future herd investment.     However, gains in productivity will recover that investment the first week she milks—while the incremental value of that cow as a dam of future replacements (the “mitochondrial” effect) is infinitely greater.     

Two of these bulls (Apple Jack Red *CV and Popular *HH2) carry lethal recessives:  
CVM*      more stillborn, deformed calves if you have carriers from Elton, Convincer, Durham, etc
HH2C*     more repeats due to failed conception if you have carriers from the Comestar Laurie line.
By contrast, neither “Bence” nor “Ernesto” (relatively outcross bulls) carry any known lethals.


Look across your pens of heifer calves, of yearlings to breed, of springers to calve.   The decision to make them better than their dams (your current milking herd) was made last year or the years before.     How much thought and effort went into making them better?

As commodity producers (commercial milk, non-specialty beef) profits flow to the least cost producer.    This is why, in recent years, the emphasis in dairy genetics shifted from a single-trait focus on “milk” to a multi trait focus including calving ease, fertility, cell counts, and longevity.    

As dairy managers, there is also a trend toward increased interest in breeding programs that help to produce more uniform, more physically adaptable replacements.    It doesn’t just matter which bulls you use, it matters when (on which cows) you use them.

Our program is in step with these two trends.    Inside is a description of how to use it to gain the most in every new heifer you raise.

Mich Livestock Service, Inc        PO Box 661       Ovid,  MI  48866     ph (800) 359-1693 toll free

Wednesday, February 10, 2016


From the September 2013 Dairy Route letter

In the beginning, we sorted breeds into “dairy” breeds and “beef” breeds, based on whether they turned feed into milk or weight gain.     Then evolved defining the “dairy” cow as milking a 305d lactation, and the “beef” cow as drying up after weaning a calf (calving every year, in either case, with the new grass).

We measure average cow size.    The bigger cow is “less” efficient, the smaller cow is “more” efficient.   (This is the New Zealand BW model.)     In other cases we do not even bother to measure size, we just will impute it from stature.   The tall cow is assumed bigger thus less efficient, therefore the shorter cow is more efficient.    (This is the USA Net Merit model.)

In the abstract, these approximations seem more true than false.    They are also virtually meaningless, as no feed has been put in front of any of these cows yet, to see what they do with it.

It depends.

The NorthCoast Group [Jersey breeders in Ohio] sets their genetic model for optimal production as:
15 times body weight annually of milk volume testing 5%bf and 4%pr calving annually over a 14 year lifetime (12 calvings, 12 lactations).    The feed efficiency is thus predicated on “milk x body weight”.      An 800 pound cow needs to produce at least 12,000 pounds, a 1000 pound cow at least 15,000 pounds, a 1200 pound cow at least 18000 pounds… (testing like a real Jersey).    The record holding Jersey cows have produced over 30x body weight (in individual record lactations).     More realistic, the lifetime record Jersey cows have produced 20x body weight over all their lactations.  

Many feed nutritionists compare cows on an “Energy Corrected Milk” basis.     Roughly, a Jersey cow milking 50 pounds daily with 5%bf and 3.7%pr equals the caloric value of a Holstein cow milking 80 pounds daily testing 3.5%bf and 3.1%pr.       Thus, any “efficiency” measuring must reflect the value of the milk, not just the gross volume of fluid produced, against the cost of the feeds consumed.

This is the next step in genomics?

Most population geneticists remain excited by Genomics because they wish to figure out the “genetics” of feed efficiency.     They want to locate the “f/e” genes, so they can feed it into their index models and tell us who is the “perfect” bull to sire “it”.

So far, this has not happened.    Our DHI testing systems collect data for herd averages and bull proofs, but they really do not measure every trait needed for describing phenotypes on relative efficiency.

Meanwhile, the latest generation of animals most favored by Genomic ranking are being criticized in some circles for actually lacking competitive growth rates, which most of us would see as a starting point for assuming “feed efficiency” genes are present and acting.

How much “milk” do we produce per acre?

The most basic ration assumption is --- one pound of dry matter intake produces one pound of milk.   The more feed the cow eats, consistently, day after day after day, the more likely she will milk.    
Our summer newsletter showed various grass forage varieties that produce from 3.5 to 6 tons D/M per acre (7000 to 12000 pounds of milk per acre, IF totally digestible forage energy).      
How many acres does it take to feed a cow?

On the above basis, you might assume you need two acres per cow, IF she would eat the volume of forages implied:    (14000 pounds = 45.9 pounds DM/day; 21000 pounds = 68.85 pounds DM/day)

Interestingly, a typical 1000 pound Jersey will make 14000 pounds, and a typical 1500 pound Holstein will make 21000 pounds, because they would eat 45 or 68 pounds of feed daily if offered.     50% more size = 50% more milk volume.     Thus size and average milk yield have near linear relationships.

To push production beyond the physical limitations of cow appetite and functional body capacity, we have one genetic option (faster metabolism) and one feed preparation option (more energy density).   It is in the development of “Total Mixed Rations” that first, we figured out how to get cows to eat stuffs she would not eat by itself, and then, how to process supplements into more energy-dense particle sizes.

This is the real reason why the conventional (TMR) industry argues their approach is more efficient—it is a “production per cow” capability that cannot be matched in a rotational grazing system.   However, the grazing community can argue that they may harvest more pounds of milk per acre than is possible under higher cost mechanized harvesting.     Neither has answered the question of “feed efficiency” but I would argue it is more important to answer the question of farm profitability first, of which gene choices is just part of the total management puzzle we must assemble.

Impact of assumptions over time    
Does it imply some eventual stupidity if we continue to expect new heifers to milk more than their dams but we follow genetic models that make the heifers smaller than the dams?    Perhaps.     It is more true that in the AI era, we have made small breeds taller, big breeds leaner, by skewing type selection toward models of faster maturing individuals found by production breeders and their AI sire partners.   We also tended to favor milk volume over component % density (creating the illusion of more feed efficiency) as well as extended lactation over reproductive regularity (delaying added energy demand for rebreeding)-- until the milk market reality penalized that stupidity with the multiple component pricing program, and later cow markets penalized low reproductive rates with higher replacement cow prices.

In any biological system, each choice creates a multiple of possible outcomes
What we must conclude is this – the feed efficient cow is not the tallest, or shortest – is not the biggest or smallest – is not the highest producing or high milk price generating –

As a calf, she must live.     As a baby, she must thrive.    As a weaned adolescent, she must grow on the available forages and reach puberty on time to stay in her group.     As a yearling she must conceive as a result of visible cycling and receptive breeding.     She must calve unassisted or risk severe setbacks in first lactation that could end her life prematurely.    She must nurture her calf to life.    She must eat and come into competitive production while also succeeding at postpartum recovery.      Next she must be returning to cycling so as to conceive again as required under the management design.   She must stay healthy given the impact of systemic health on milk quality measures.    She must finish growing into the mature physique dictated by her genotype.    If successful on all these energy driven functions, she will attain longevity as a productive herd member and produce genetically superior replacements.    In this appraisal, I continue to believe that feed efficiency is implied genetically in proven longevity.

Is “feed efficiency” more important than “longevity” ??

We are very aware of the next focus of Genomics—it is going to be “feed efficiency” because of the industry’s ongoing concerns over feed cost vs milk price ratios.

I would be the first to agree that feed costs must be under control (it remains a major reason why we converted our dairy to rotation grazing a dozen years ago).    I have become a bit slower to assume that population genetics will give us the right answers.

The dairy cow, unlike any other farm animal measured, produces a multiple of energy driven products and required functions.    To measure the genes of each of these on an individual basis will be successful.    To prioritize them in selection, when all will be needed in any functioning animal, will be more difficult.

We may already have the gene selection and mating selection processes that will lead us to profitability over current feed costs—we just have to step back and see the big picture of what has to fit together, and what the real limitations of land acreage base and forage production of NDFd from that base mean for our production targets.

Sunday, February 7, 2016


From the August 2013 Dairy Route Letter

 I was in Ontario on an “aAa” tour and the same subject came up twice.     First a young Dutch couple and next an experienced Canadian breeder commented on the strength of the cow market.     Quota was being assigned for more fall milk needed in the Ontario market, and dairies were adding summer-fall calving cows, at prices from $2500 to $3000 per cow.

“Why the $500 differential?” I asked.     Was it related to age? To pedigree?  Genetic value or Genomic testing?    None of the above.    It was totally visual.     “When the fancy one came in the ring, bidding picks up, more interested buyers, and you earn the higher price” so says Rene Strick of Woodstock.   It was the same with John Dortman of Strathroy, who was selling fresh cows he purchased as thin heifers last season to help neighbors short of feed.     “I have handled sale animals for years and the market has come back to buying with their eye.     All the genetics on paper is not keeping these commercial barns full.     They will pay more for better type, because the commercial type has not lasted for them.”

What is the eyeball difference between “commercial type” and “better type”?

(1)   Balanced udders.     Not big full rear udders, not bulgy fore udders, but an evenly formed and snug udder, with teats that hang plumb and central to the quarters—a visually milkable udder, up out of the way of harm when feeding or lying in free stalls, avoiding the meatiness that makes them age faster.

(2)   Good bodies.       Not shallow, not narrow, not slabby ribbed, but expressing some capacity for the forage content of the modern lower cost ration.      Proportional in length, width and depth to fit within a new standard free stall.     Open and elastic in the rear rib and flank to aid in persistency.    1300 pound heifers that become 1700 pound cows.

(3)   Feminine with style.      A cow should not look like a steer or a race horse.    She should have a bit of feminine refinement in her head, neck and bone quality.   She needs a wedgy shaped body (a “sharp” shoulder with broad chest, ribs angle back deeper toward the belly, widen from front end to a broadly placed hip, pelvic structure open enough to house the udder as it matures).

(4)   Fluid motion legs.       Neither “curved” nor “straight”, but erect enough to hold skeleton up and springy enough in joints to avoid crampiness.      They need some substance of bone to support weight and they need to be positioned properly, both rear AND front legs, to carry the load evenly and prevent uneven hoof wearing.     A sturdy stance with a straight tracking in the front end.

(5)   Well blended.       First, the sort of appetite that helps a cow avoid a persistent negative energy state.   Second, enough muscle cover over bones to not only aid in mobility but to cushion bones from injury.   

Wednesday, February 3, 2016

So what is all the noise over polled Jerseys in the USA?

Greg Palen           July 2013

A polled heifer calf tops the Ohio Spring Production sale, selling for $14,000 (US).    Not more than a month later, a polled heifer tops the National Heifer Sale selling for $26,000.   
All three polled consignments exceed the sale average.    What is going on here?

Basically, the polled possibility in Jerseys is having the same expansion in the population of breeding caliber cattle as it is having in the Holstein and Red & White breeds.    This is in part driven by interest from globally marketed AI studs to provide more polled options.       But it has been building behind the scenes for many years  

As an example, there is the Northcoast Group, based in east-central Ohio, an association of Amish and Mennonite dairymen producing milk for local cheese plants, who switched to Jerseys from Holsteins in the 1970s, and by the 1990s had added “polled heads” to the list of traits and qualities they wanted their Jerseys to have.    This group has been able to privately market semen across the country from their polled Jersey sires, mostly among a clientele involved in grass based and certified organic milk production.     They have not sold any polled bulls to AI studs, having a “cow line” pedigree and phenotypic selection philosophy, but dissemination of their cattle around the country has influenced the rising interest in naturally hornless Jerseys for both purebred and crossbred use especially since they found their first homozygous (“pure polled”) sire over a decade ago.

Polled heads is the sort of “dominant” (single parent) gene trait that attracts the interest of the “dirt farmer” dairyman—the highly practical farmer putting his crops through cows as a marketing strategy, who wants no nonsense, easy keeping, highly adaptable cattle.   The handful of breeders who kept track of breed sources of polled and multiplied their polled numbers, were not chasing indexes—they were chasing combinations of practical (health and fertility) traits they could align with the polled gene and produce seedstock for their dirt farmer neighbors’ use.    

Thus polled is a trait that grew “under the radar” until the commercial demand for polled heads could no longer be ignored by AI systems trapped in the “ranking game”.    Lack of awareness of polled gene sources led many to the misimpression that polled dairy cows must be crossbreds with beef ancestors (highly ironic in the case of Jerseys, because the original Jersey Island cows were ALL polled prior to the Enclosure Acts in the 1800s).

People in tune with market desires saw that polled sires would sell without ranking at the top of the genetic lists, which for too long were narrowly defined by “milk” and “type” alone, a focus narrow enough to produce a shrinking gene base in all major breeds.    One of the overlooked factors in polled dairy cattle is that they represent a useful “outcross” to the popular genetic lines.      Taurus Service in Mehoopany, Pennsylvania, a regional AI system adept at focusing on market niches, began to specialize in bringing polled dairy sires to the broad AI market, and now boasts an ongoing program of polled Holstein, Red Holstein, Jersey and Milking Shorthorn bulls as part of their “New Age” alternative genes venture (with Genetics Australia).

Every major AI system in the US now actively seeks polled Jersey and Holstein sires.   
Their stimulus was the exploding demand in Europe for polled Holsteins, a result of the animal rights initiatives that strongly affect animal agriculture across the EEC (dehorning according to regulations now costs the European dairyman E70 per horned calf!).    But in the case of Jerseys and polled, the USA motivation relates more to the recognition that Jersey is the one major breed with a strong growth trend, and that polled is in demand in the same sort of geography where Jerseys are in demand—grass based dairy sectors, more semitropical climates, seasonal calving management, and large scale dairying with premium costs for purchased feed (affecting the supply of heifer replacements).    

The advantages of polled show up faster as herds get larger.   Dairy labor is at a premium, and dehorning is nearly the nastiest job on the herdsman’s list.    Dehorned calves lose a months’ growth alongside their polled stablemates, proving there is a health reaction to dehorning no matter what technique is used.    Where dehorning leaves wounds, parasites find entry to the calf’s head and force us to expenses of individual treatment, may cause disease or death if there is a shortage of attention to this age group.    The odd cow with horns in a mostly dehorned herd is a constant risk of injury to her stablemates.

Thus the market value of naturally polled genes is rising as we better recognize the costs of dairying can rise faster than our incomes.     Better managers know their costs and seek those that can be eliminated without affecting the income flows and profitability.    Polled as a genetic option has actually made “genetics” more interesting to expansion dairymen who were increasingly sensing the index ranking game was a matured technology.

In our herd, starting twelve years ago to introduce the polled gene into our Jerseys, we are now at a point where we only keep polled replacements.    We have found it easy to breed the kind of cows we wanted from the limited polled sources, utilizing a practice of mating our favored cows to a phenotypically compatible polled bull, saving any polled bulls born and using them for the following generations where we felt they fit.    Over these recent years, mating a third of our herd to our own and Northcoast Group bulls, and occasional use of AI source polled bulls, utilizing the “aAa” Breeding Guide (Weeks’ Analysis) to determine the final  mating, we now find that two thirds of our herd is polled cows, and the fertility and longevity of these cows insures we only need 20% of our annual heifer crop as replacements.     Again, although a “stand alone” gene in nature, the pioneering breeders of polled Jersey cattle were breeding for their own profits rather than AI market popularity, thus favored “complete cows” (easy calving, fertile, healthy, long lived) as the dams of their polled sires, and these genes for superior strength of constitution show up as you multiply the polled effect over additional generations.

Our heifer raiser (who handles the marketing of all our surplus heifers) had a phone call this spring from a prior purchaser, who said “I want first chance at every heifer you will calve in 2014” to fill his recently building dairy facility.    “Your heifers get better each year they calve, while the others we have bought just wear out too early.”     What made our heifers more adaptable to his needs as a new dairyman?     The bundle of selection traits and mating qualities we have assembled from polled Jersey sources, essentially an outcross to the mainstream of prior popular Jersey focuses.     Anyone with Jerseys now has this opportunity available to them, across the Jersey world.

(Who is Greg Palen?)

Owner of Netherhall Dairy Farm (rotational grazed pasturage, forage based feeding) and Netherhall Polled Jerseys (selected for cheese yield component % levels and longevity).
This dairy began in 1979 with purebred Holsteins, added Jerseys in 1986, and became all Jersey in 1994 with a switch from TMR feeding to seasonal rotation grazing.     Current management of the farm and herd is the responsibility of Tim and Liz Heinze.

Owner of Mich Livestock Service, Inc, an independent AI service based in and serving central Michigan, acting in distribution for smaller AI systems like Taurus Service, Inc and its affiliates, International Protein Sires, Trans World Genetics, featuring polled sires.
This business was begun in 1952 by his father, Charles, with Curtiss Breeding Service, and has three salesmen, including the recent addition of a full line of forage seed species.

Acted in wholesale distribution and dealer recruitment for Tri State Breeders Cooperative (now called Accelerated Genetics) for Michigan, Indiana and Ohio from 1980 to 1990, and then performed the same function for Semex Canada from 1991 to 1995 in the same states.      Both companies now service these areas with exclusive employees.

Approved to provide “aAa” Breeding Guide (Weeks’ Analysis) in 1994, now with clients in Michigan, Ohio, Kentucky, Indiana, Pennsylvania (USA) and Ontario (Canada).    This is my current business focus.     “aAa” acts independently of any AI stud or herdbook and provides analysis of qualitative gene effects, distinct from linear trait evaluation methods.

Graduate of the University of Michigan School of Business Administration in 1975 with a major emphasis in marketing and a minor in finance.     Married 37 years to Wife Sue, three grown children, one grandson.   

President of the Ovid village Downtown Development Authority since 2004.     Elected Supervisor of Ovid Township in 2012.      Plays tuba in “Ovid-Elsie Community Alumni Band” and cuts firewood all winter for exercise.