Jim Eschenbaum is frustrated about the lack of forage testing accuracy and consistency among labs — even labs certified for accuracy by the National Forage Testing Association.

Eschenbaum, a Miller, SD, hay grower, says varying results cost him one customer — and almost his reputation.

“The problem originally arose with a dairy I agreed to send hay to at 180 relative feed value (RFV) or above. The original feed test was 193. They rechecked the hay because they didn't feel the cows were milking as well as they should have been if the hay was 180 or above.”

The dairy had a new sample of hay tested at Dairyland Labs — a lab different from the one Eschenbaum had sent the original sample to. Its result came in at 160 RFV — 33 points below Eschenbaum's 193 RFV.

“The first thing my customer thought was that I'm taking erroneous samples to try to get my RFV up good enough to sell it to him at 180 or above,” says the grower, who harvested hay from 1,000 acres this year. “You end up being the questionable hay seller every time you run into differences between labs.”

The fact is, there will always be variability when evaluating forages, says Dave Mertens, dairy scientist with the U.S. Dairy Forage Research Center, Madison, WI.

Some variation is caused by how samples are processed and analyzed and some by the way samples are taken. A certain amount of variation is unavoidable, says Mertens, who is also a driving force behind the National Forage Testing Association (NFTA).

“The goal of the organization is to get better repeatability within and among forage-test laboratories,” he says.

Formed 22 years ago, NFTA set up a volunteer governing board made up of representatives of forage testing labs, the American Forage and Grassland Council, the National Hay Association and USDA-ARS and university scientists. Its certification program evaluates labs on their performance. This year, about 150 labs were certified, which means they earned “C” grades or higher based on accuracy for dry matter, crude protein and ADF and NDF analyses averaged across six samples. Certified labs can then advertise as such, carrying the NFTA logo in literature, for example.

Although NFTA has helped standardize forage testing, Mertens says, the organization wants to improve its proficiency certification program and has been going through an evaluation of its own.

Currently, NFTA sends participating labs four alfalfa samples as well as one corn silage and one grass hay sample. Samples are first ground so each lab gets a representative sample that lets NFTA compare its results to a predetermined average.

“The current proficiency certification has two shortcomings. One, we don't evaluate the sample preparation part of the analytical procedure in the lab, because when a farmer or any client sends a sample in, that sample has to be dried and ground before it is analyzed.

“The other thing is, when a sample comes in ground and identified as an NFTA sample, we tell labs they're supposed to treat the sample like any other. But they are always aware that sample is used for certification.”

When an NFTA sample appears at a lab, says Tom Keene, lab technicians are careful to follow the instructions, or reference methods, provided by the organization because they want to get passing grades. But unground samples from growers or dairymen aren't always given the same consideration and at times are analyzed using less-expensive shortcuts, says Keene, a University of Kentucky hay marketing specialist and National Hay Association representative on the NFTA board.

At their annual meeting this past spring, NFTA board members presented a study exploring ways to improve accuracy. “We're trying to move toward samples that look like what clients send in,” Mertens says. “Eventually, there is some interest in sending blind samples out — where they wouldn't be labeled as coming from NFTA.” The organization has also been talking about increasing the number of samples it requires for certification.

“We're on the right track, but we're not there yet. We've got a lot of work left to do,” emphasizes Keene.

Growers add variability to forage results, too, says Dan Putnam, University of California-Davis extension forage agronomist who is also on the NFTA board. Those who sample entire barns of hay rather than single cuttings from single fields, probe only good-looking bales or subsample haphazardly can reduce forage-test accuracy. Putnam suggests proper sampling procedures in the accompanying story, “Sample Right To Improve Accuracy,” page 7.

Mertens adds that buyers and sellers need to be reminded that their test results are just estimates.

“You've got a stack that has 100 tons of hay in it and you're sending a sample the size of a bread wrapper. Then the lab grinds that and dries it and analyzes a sample that would fit roughly into a fourth of a teaspoon.

“Each new sample and analysis will give you a slightly different estimate of the true value of the forage. You have to realize that's just an estimate of what's in that stack. And it all depends on how well you sampled that stack, how well the lab ground and sampled your sample and how well the analytical work was done.”

Eschenbaum says he follows proper sampling techniques. He uses a sharp, ½" hay probe that cuts small cores. If a field has 100 round bales, he'll sample every 10th bale in the row to get an accurate average, no matter what the bale may look like.

“I want to sell hay as the first or second cutting and try to get a good average of what the whole field was. If there is something that I feel divides the feed quality of that field for some reason, I will split a test on a field.”

Since that ill-fated dispute over forage tests, he now sends samples to the same labs his customers rely on, usually Dairyland Labs.

Yet that, too, got him into trouble.

“I approached a dairyman milking 1,600 head and told him I had some really nice hay. He also wanted hay 180 or above and I said I had Dairyland tests that ran from 150 to 174 from several fields.”

But the dairyman wanted tests from another lab Eschenbaum was familiar with. “I said, ‘That's no problem. I can core the bales and they'll be 180 or more at that lab.' ”

He cored the same bales he'd sampled before on one field, just 3” from the previous cores, and sent those samples to the lab the dairyman specified. Dairyland results had showed an RFV of 160; the other lab, 227 — a 67 RFV point difference.

“I called the dairyman and he said, ‘I'm going to be totally honest with you. Anytime I see a feed test that's over 200, I'm a little skeptical about how samples were taken.'

“So right there, I had another potential customer questioning my integrity,” says Eschenbaum. “But it ended okay. He said, ‘Send me one load of hay. I'll take a dozen loads if it's like what you say it is.' ”

The dairyman bought 11 more loads and would have taken more if Eschenbaum had had it.

In reality, says Mertens, Eschenbaum took different samples at different times — and although differences in RFV that big aren't the norm, they happen.

“If the samples are more variable, then we would expect those labs to have greater variation between them even if they did the same thing exactly the same way perfectly. Their two samples didn't start out to be the same.”

He recommends, when there is a dispute over forage-test numbers, that ground samples from each lab be sent to the other for a recheck. That way both labs can test both samples and a grower can decide if the samples or the analysis was the problem. If the two labs agree on each sample, the samples are different from each other. If the labs disagree on both samples, their analyses are different.

Eschenbaum got similar advice from a lab manager when he told him he was planning to core bales and send samples to three labs to compare results.

“He said, ‘If you're going to do that, each lab is going to be working with a different hay sample. We're not going to have comparable results.' ” The manager offered to grind and divide the samples, then send them back to Eschenbaum to send to whichever labs he wanted.

He sent the ground samples to three. RFV at lab A was 159; at lab B, 180; and at lab C, 189.

A 30-point RFV spread is significant, he says. “There are guys who are pricing their hay off of RFV, some at 50, 60 or 70¢ per point of RFV. If they're pricing their hay off of RFV, well, the dairies will obviously want to use the labs that test low.”

There isn't an easy solution to the problem of inconsistent forage tests, in part because of the nature of the crop, says Putnam.

“We're talking about a very heterogeneous mass of plant material,” he says. Proper sampling and choosing a certified lab are very important to getting good test results, he adds.

Growers can also reduce disputes by testing at the labs preferred by their dairy clients, Eschenbaum and Mertens agree.

And NFTA-certified labs need to analyze all samples as well as they do NFTA's, Keene stresses.

Yet dairymen, growers and sellers all need to remember that there will always be variation among forage tests. And that test results are single estimates of what's in a stack of hay, Mertens says.

The best way to avoid disputes and estimate the true value of a stack of hay is to average the results of several samples from several labs, Mertens says. Eschenbaum, he adds, could have averaged the three labs' RFV numbers (159, 180 and 189) rather than picking just one result.

“We have to get away from the idea that any single analysis will be the correct one and that all others are wrong. There will always be some differences between estimates, even when sampling and analyses are done perfectly,” Mertens concludes.

Sample Right To Improve Accuracy

Growers can reduce the chances of inaccurate forage tests by providing samples representative of the stack or field to be tested, says Dan Putnam, University of California extension forage agronomist.

“The most frequent difficulties with hay quality analysis can usually be traced to sampling problems,” he says. “The first step is to make sure you test a single cutting from a single year and a single field. If you decide you're going to combine four or five fields, you're going to introduce lots of variation.”

Putnam says growers should sample less than 200 tons at a time and as close to feeding or point of sale as possible.

“One of the common mistakes a person makes is not having the right coring device. We recommend a smaller coring device with a very sharp tip.” Big probes give too much material and corers that are too small don't adequately represent variation in a stack, he says. Probes should be ⅜-¾" in diameter, which allows the advised 20 cores per sample and a sample of about ½ lb.

Sample randomly — don't probe just the best bales, but try to represent all areas of a stack, Putnam says. Sample ends of square bales, between strings or wires and not near the top or bottom. Probe at a 90° angle 12-24" deep. “Round bales have a slightly different technique because you have to go toward the center of them.”

Seal your composite sample in a plastic bag, protect it from heat and send it to an NFTA-certified lab as soon as possible.

Don't send in flakes or “grab samples.” They do not represent the quality of a stack of hay and can be too much of a sample for a lab to handle economically.

Splitting unground samples to compare labs doesn't work, Putnam says. To test performance among labs, ask for a ground sample back, then split it by carefully rolling it on paper and, using a spatula, divide it as you would a pie. Be careful not to allow “fines” to segregate out of the samples.

Putnam says that growers or buyers who want to improve their sampling methods or anticipate price or quality arguments, should “certify” their methods of sampling. Sampling protocols and a 30-question sampling certification exam are available at www.foragetesting.org. Listings of available hay probes and certified samplers can also be found on the Web site.