Introduction

There are four natural resources required by honeybees for survival: water, resin, nectar, and pollen (Seeley 1985). Water is used to cool the hive and to dilute the honey fed to the larvae. Resin is utilized to reinforce the hive, seal off decaying wood, and plug up holes. Nectar is the major source of carbohydrates from which honeybees obtain their energy. Nectar is collected by foraging worker bees and is carried back to the hive in their honey stomachs. Upon returning to their hive, the nectar is usually transferred to hive workers for processing into honey, although it can be fed directly to the brood or to the adults (Winston 1987). Enzymes from the bee's hypopharyngeal glands are added to the nectar in the bee's crop. These enzymes break down the nectar into simple forms of sugars, which are easier for the bees to digest. These enzymes, in addition to the high sugar content, also protect the stored honey from bacteria. The water in the nectar is then evaporated off of the worker's tongue. The nectar is placed into cells and fanned to further reduce water in it. Through this process, the water content in the nectar is reduced to less than 18% (Winston 1987). Once the evaporation process is complete, the nectar is considered "ripened" and is called honey. The cell is capped with wax until the honey is needed for feeding to the larvae or the adults.

Some statistics about honeybees and the production of honey are important to note. Years of observation and research have revealed a number of facets about this subject. For example, it is estimated that to make one pound of honey, honeybees must visit about two million flowers, fly a total of about 50,000 miles, and carry about 37,000 loads of nectar back to the hive. According to the British entomologist, Arthur Thomson, during the main flower blooming periods it is common for the bees from a single hive to visit as many as 250,000 flowers during the course of a single day (Teale 1942). Some flowers, such as the ones of a tulip tree (Liriodendron tulipifera) each produces about a teaspoon of nectar. Other flowers, such as the ones of white clover (Trifolium repens), produce only enough nectar to cover 1/20 of a pinhead (Crane 1975).

Each worker bee is able to carry a load of nectar equal to one-half its total weight and during her lifetime one worker will collect enough nectar to produce about 1/12 of a teaspoon of honey. During nectar gathering, a honeybee consumes 0.5 mg of ripe honey per kilometer of flight. To produce one liter of surplus honey the worker bees of a hive will consume eight additional liters of ripe honey as food. Ripened honey and pollen stored in a hive are the food sources eaten by the bees. Feeding a bee larva from the egg to maturity requires about 142 mg of honey (Winston 1987).

Where does the pollen in honey come from?

Pollen is the bee's major source of proteins, fatty substances, minerals, and vitamins (Gary 1975). It is essential for the growth of larvae and young adult bees (Dietz 1975). Honeybees remove pollen from an anther by using their tongue and mandibles. While crawling over flowers, pollen adheres to their "hairy" legs and body. The honeybee combs pollen from her head, body, and forward appendages, mixes it with pollen from her mouth, and transfers it to the corbicula, or "pollen basket", on her posterior pair of legs. When "loaded" with pollen, she will return to her hive. Once at the hive, workers pack the pollen into the comb. To prevent bacterial growth and delay pollen germination, a phytocidal acid is added to the pollen as it is packed into the comb. Other enzymes produced by worker bees are also added which prevent anaerobic metabolism and fermentation thereby enhancing the longevity of the stored pollen. Once completely processed for storage, the pollen comb is referred to as "bee bread" and is ready for later consumption by the bees. The protein source needed for rearing one worker bee from larval to adult stage requires approximately 120 to 145 mg of pollen (Alfonsus 1933; Haydak 1935). An average size bee colony will collect about 20 to 57 kg of pollen a year (Armbruster 1921; Eckert 1942). In most cases the primary foraging areas for pollen are the various insect-pollinated plants bees visit for nectar. However, honeybees will also visit a number of species of wind-pollinated plants for which their only purpose is to collect pollen. Wind pollinated species of Salix (willow), Quercus (oak), Celtis (hackberry), and many species of grasses (Poaceae) as well as some of the wind-pollinated types of composites (Asteraceae) are considered important pollen sources for foraging honeybees.

Melissopalynology is the study of pollen in honey. For over 100 years the literature pertaining to the study of pollen in honey has been termed or spelled several ways, including: mellissopalynology, mellittopalynology, and melittopalynology. According to Paxton's Botanical Dictionary (1868), both "melissa" and "melitta" mean "a bee". The scientific name of the honeybee is Apis mellifera L. The word "melliferous" comes from the Latin word mellifer (honey) and the suffix -ous meaning "having, full of, or characterized by". The International Commission for Bee Research uses "melissopalynology", which is the term used throughout this essay.

Pollen can be incorporated into the honey produced in a beehive in a number of ways. When a honeybee lands on a flower in search of nectar, some of the flower's pollen is dislodged and falls into the nectar that is sucked up by the bee and stored in her stomach. At the same time, other pollen grains often attach themselves to the hairs, legs, antenna, and even the eyes of visiting bees. Later, some of the pollen that was sucked into her stomach with the nectar will be regurgitated with the collected nectar and deposited into open comb cells of the hive. While still in the hive the same honeybee might groom her body in an effort to remove entangled pollen on her hairs. During that process pollen can fall into open comb cells or the pollen can fall onto areas of the hive where other bees may track it into regions of the hive where unripe honey is still exposed in open comb cells. Some worker bees also collect pollen for the hive. The smooth, slightly concave, outer surfaces of the hind tibia in worker bees are fringed with long hairs that curve over the tibia surface to form a hollow area. This hollow area is called the "pollen basket" or orbicular. The worker bees collect pollen with their front and middle legs and then deposit it in their cubicula (Snodgrass and Erickson 1992). In the process of depositing collected pollen into special comb cells some of it can fall into the hive or into open honeycombs. It is also noted that occasionally worker bees might add pollen to the nectar they are transforming into honey.

Airborne pollen is another potential source of pollen in honey. Many types of airborne pollen produced mostly by wind-pollinated plants that are not usually visited by honeybees can enter a hive on wind currents. These airborne pollen grains are usually few in number, when compared to the pollen carried into the hive by worker bees, nevertheless, those pollen types regularly enter a hive on air currents and can settle out in areas where open comb cells are being filled with nectar. Sometimes airborne pollen is deposited into ripened honey when it is being removed from a hive by the beekeeper. Although the pollen rain for various regions consists mainly of airborne pollen, and those data are often used in forensics, archaeology, and ecology to identify a specific geographic region, those pollen data are not always as useful in melissopalynology because they generally form only a minor (?) fraction of the total pollen spectrum found in a honey sample.

Pollen is an essential tool in the analyses of honey. Taxa of pollen are used to indicate the floral nectar sources utilized by bees to produce honey (Lieux 1975, 1977, 1978; Moar 1985; Louveaux et al. 1970; Sawyer 1988; Van der Ham et al. 1999). Thus, the relative pollen frequency is often used to verify and label a honey sample as to the major and minor nectar sources. This information has important commercial value because honey made from some plants commands a premium price (i.e., sourwood, tupelo, buckwheat, or citrus honey). Even non-premium grades of honey require certain types of verification because they must be correctly labeled before being marketed. Identifying and quantifying the pollen in honey samples is one of the best ways to determine the range of nectar types used to produce a honey, and therefore label it correctly based on actual foraging resources. Another reason that pollen analyses of honey are often required is to identify the geographical source of origin. The combination of wind and insect-pollinated taxa found in a honey sample will often produce a pollen spectrum that is unique for the specific geographical region where it was produced. Because of trade agreements, import tariffs, and legal trade restrictions, most of the leading honey-producing nations of the world require accurate labeling of honey before it can be sold. This is especially true for the EEU that has had strict labeling regulations for honey products since 1974 (EEU 2001).

The history of melissopalynology: a brief overview

At the end of the nineteenth century, Pfister (1895) examined the pollen contents of various Swiss, French, and other European honeys. Through his analysis, he demonstrated the possibility of determining the geographical origin of honey from the pollen within it. He was able to identify many of the pollen grains he found because of earlier studies of pollen morphology, structure, and identification of European pollen types by botanists including Guillemin in 1825, Fritsche in 1832, Mohl in 1834, and Fischer in 1890 (Wodehouse 1935).

The history of the scientific investigation of U.S. honey begins in the early 1900s when a researcher working for the United States Department of Agriculture (USDA), W. J. Young, published a brief report on the analysis of domestic honey produced in the United States (Young 1908). One of the reasons that Young says he examined the pollen content of honey was to determine if pollen studies could be used in the future to "judge the adulteration of a sample" (Young 1908). His hypothesis was that if honey was adulterated with sugar syrup, this could be detected by finding a reduction in contents of the pollen. However, the method he used to determine the pollen concentration values of his samples is not considered accurate by today's standards. Currently, most melissopalynologists use larger amounts of a honey sample (10 g) and ratios of pollen grains per gram of honey based on counts derived from comparing the pollen to percentages of introduced "tracer" spores added to each sample before analysis. Young, on the other hand, determined his published pollen concentration ratios for 19 of his 100 honey samples by extracting only one gram of honey from each sample, diluting it with water, and then counting a small portion. He doesn't explain why he did not attempt pollen concentration studies for the other 81 samples in his study. When deriving his concentration values, Young relied on identifications of pollen types that still contained their pollen's cytoplasm, waxes, or surface lipids. By failing to process the honey samples in order to remove those components from the pollen, Young's ability to make precise identifications of pollen types was probably difficult. Young states that he used the pollen concentration value for each of his 19 samples as a basis for predicting the expected pollen concentration values for future studies of each type. Based on his work, Young determined that the range of pollen concentration values varied from a low of 123 pollen grains/g to a high of 5,410 grains/g of honey.

The second reason Young examined the pollen contents of his 100 samples was to determine the identity of "the pollen from a large number of flowers known or suspected to be visited by bees in different sections of the country" (Young 1908). Ninety of the 100 honey samples had a purported source that was provided by the beekeeper, such as "melon honey, clover honey, or cotton honey", etc. The other 10 honey samples were listed as being from mixed floral sources. Thus, for each of the 100 samples the major pollen types were compared to see if they corresponded with the suspected honey source. For example, each of his pollen analyses reported first the sample number followed by the type of purported honey based on the report provided by the beekeeper that collected the sample. Next, Young listed the state in which the honey was collected and then the pollen types he found in the sample. Each of his entries is listed such as this one: "sample 61. Melon (Illinois). Clover, Melon, Cruciferous, Polygon um, Alfalfa, Basswood, Composite (two kinds), and Ellipsoidal types". Unfortunately, Young provides no explanation as to how many pollen grains he counted for each sample, nor the percentage of each pollen type he identified. He also does not say if the pollen list for each sample is based on the descending order of pollen frequency of each type found in the sample.

Although Young's report focused mainly on the chemical aspects of honey and honeydew samples, he was one of the first to examine the pollen contents of honey. He made a key to the pollen grains commonly found in U.S. honey, discussed the importance of protecting honey samples from airborne contaminants, and discussed the various kinds of structures (insect parts, fragments of the comb, fungal spores, dust, pollen, etc.) that are likely to be encountered when examining honey. Nevertheless, his actual pollen data are of little research value to melissopalynologists today. Unfortunately, Young provides no explanation as to how many pollen grains he counted for each sample, what was the relative percentage of each pollen type he listed, or even which pollen types were the most or least important in each sample.

In 1911, Fehlman published his work on the pollen spectra found in various examples of Swiss honey (Maurizio 1951; Maurizio and Louveaux 1965; Lieux 1969). Fehlman's work was significant because he was the first European to use pollen as a way to identify and differentiate honeydew from nectar honeys, and to demonstrate that pollen contents were the key to determining the nectar sources in honey samples.

In the United States during the 1920s Parker (1923) conducted a study of bees and the honey they collect. His research remains an important contribution for a number of reasons. First, he described 28 different kinds of pollen collected by honeybees, and included photographs of the 12 most important ones. Second, like others before him, Parker was convinced that the pollen content in honey was a valuable tool for identifying the foraging sources used to make it. Third, he recognized that if bees were trapped on their return to the hive, the pollen recovered from the nectar in their honey stomach would identify the foraging areas being utilized by the hive.

Other research advancements in melissopalynology during the 1920s were made by Betts and Allen, who worked separately on English honey. Betts (1923, 1925) made sketches of 15 different kinds of pollen sources found in English honey types and she suggested that flowers from herbarium specimens could be used as a source of pollen to make comparative reference samples. These taxonomically-correct pollen reference samples, she reasoned, would speed the identification of unknown types found in English honey and it would also enable researchers to add another level of precision to their identification of pollen recovered in honey samples. A few years later, Allen (1928a) noted that some pollen grains remain on the surface of the honey, instead of becoming mixed with the honey like other types of pollen. Allen reasoned that some pollen grains "floated" on the surface of honey because they must be lighter and less dense than the honey. He was also the first to report that pollen found mixed with nectar could come from sources other than the nectar plant's own anthers and pollen (1928b). For example, Allen observed that accidental contamination of nectar and eventually the honey it was used to produce could occur through several means. First, he noticed that bees that had visited one type of flower might move to flowers of a different plant species in search of new nectar sources. If that occurred, then pollen adhering to the body of the bees could accidentally fall into, and thereby contaminate the nectar of the second flower type with pollen from previously visited plants. Allen also noted that airborne pollen could easily contaminate honey when combs were being removed from hives and also during the subsequent honey extraction process. Regardless of the causes and types of pollen contamination, however, Allen reasoned that contamination was usually a minor problem and that pollen in honey mostly reflected the actual floral sources used to make the honey.

By the end of the 1920s Allen (1929) was focusing on some of the problems of conducting accurate melissopalynology analyses. He was the first researcher to caution about some of the pitfalls and difficulties of pollen identification in melissopalynology. For example, Allen noticed that dried pollen on herbarium sheets and fresh pollen collected in flower anthers looked very different from the pollen he recovered in honey. As a result of his observations, he was the first to question the accuracy of pollen identifications reported from other previous melissopalynology studies. Allen published a series of articles in Bee World (1928a, 1938b, 1928c, 1928d; 1929) where he illustrated that many of the pollen types found in honey samples look nearly identical and that various pollen genera could easily be mistaken for other pollen types because many of them looked superficially similar. Finally, he noted that because the predominant types of mounting media fixed both pollen reference material and pollen recovered from honey samples in permanent positions on a microscope slide, the pollen grains could not be rolled over in order to search for a critical aperture or morphological feature that would confirm the type's true identity. The final contribution of Allen's work (1928a, 1938b, 1928c, 1928d; 1929) was his proposed pollen classification system for English honey. In one article, for example, he cautioned that, "one should doubt the origin of a honey sample as being English if the sample contains six-grooved pollen grains". Today, we know that those pollen types are common among the genera in the family Lamiaceae and include taxa such as Mentha (mint), Thymus (thyme), and Salvia (sage). Introduced species of many of those plants now grow in English gardens, thus Allen's initial conclusions would no longer be valid.

During the 1930s and 1940s one name stands out as being the leader in melissopalynology research. Zander's (1935, 1937, 1941, 1949, 1951) five-volume work, published over a span of nearly two decades, laid the foundation for melissopalynology research in Europe. In his various analyses and reports he includes descriptions, drawings, and photographs of pollen that he found in various types of European honey. He also includes several studies of other types of material that are sometimes recovered in honey, such as fungal spores and hyphae. Because of his long and dedicated works in the field, Maurizio and Louveaux (1965) refer to him as the "leader in melissopalynology research in Europe".

In the United States the period of the 1930s represented a time when no research in melissopalynology was being conducted even though some additional research continued on plants used as honey sources. It was during the early 1930s that two valuable books on honey plants were first published: American Honey Plants (Pellett 1930) and Honey Plants of Iowa (Pammel et al. 1930). Later, in 1939, Oertel published the results of his seven-year study on the sources and blooming periods of plants thought to be principal honeybee nectar sources in various regions of the United States (Oertel 1939).

During the early 1940s two scientists working for the USDA in California, Frank Todd and George Vansell, began searching for the relationship and importance of pollen in honey (Todd and Vansell 1942). Their research began when they discovered that bee colonies would survive, but would not reproduce if they were fed only sugar syrup. Once pollen was added to the feeding syrup egg laying in the hive began with 12 hours. Their multi-year study represents the next major study in melissopalynology in the United States after Young's 1908 initial examination of pollen grains found in domestic honey. Todd and Vansell restricted their pollen and nectar research to plants and honey produced in California, because that is where their laboratory was located and they could get assistance in their study from experts at the University of California. The two researchers began their study by collecting and examining over 2,600 individual samples of nectar. One of their goals was to try to determine the number of pollen grains one should expect to find in 1cc of nectar from each different plant species. Next, they wanted to determine if the number of pollen grains found naturally in nectar samples matched the number of pollen grains found in the honey stomachs of bees that foraged on those same nectar types. Third, they wanted to discover how efficiently bees could remove pollen from the nectars they collected.

The Todd and Vansell (1942) study was virtually ignored when it was first published, but the importance of their work has now been recognized as significant because of the information they collected and the implications their data provided about pollen counts in honey. Nevertheless, Todd and Vansell admit that some of their research ideas came from a study in Wisconsin by Whitcomb and Wilson (1929), who had been studying dysentery in honeybees when they noticed that many of the pollen grains sucked into a bee's honey stomach along with nectar were quickly removed through a process of filtering. Whitcomb and Wilson noted that once nectar enters a bee's honey stomach it is filtered and within 10 minutes most of the pollen in the nectar is removed leaving mostly pure nectar in the honey stomach. They concluded that the ability of a bee to filter nectar in her honey stomach is one way of removing unwanted debris from nectar, such as pollen and fungal spores, which might germinate and spoil future honey made from the gathered nectar.

The honeybee's filtering process, as described by Snodgrass and Erickson (1992) is rapid and effective. The bee sucks nectar into a slender tube that ends in the bee's abdomen where it becomes an enlarged thin-walled sac called the honey stomach. This thin-walled sac is greatly distensible and can expand to hold large amounts of nectar. Once in the honey stomach, the nectar flows over the proventriculus which serves as a regulatory apparatus that filters and controls the entrance of food into the bee's stomach. The anterior end of the proventriculus, also called the honey stopper, projects into the bee's honey stomach like the neck of a bottle and at its anterior end is an x-shaped opening consisting of four, thick, triangular-shaped, muscle-controlled lips. The nectar in the honey stomach is drawn back and forth into the funnel-shaped proventriculus where it is filtered to remove debris such as pollen grains and the fungal spores of foul brood. The posterior end of the proventriculus extends into the anterior end of the ventriculus that is part of the bee's alimentary canal (mid gut) where digestion and food absorption occurs. A valve at the bottom of the proventriculus prevents the filtered nectar from entering the bee's digestive system and it ensures that the nectar is returned to the honey stomach. However, this same valve will open to allow debris removed from the nectar to pass into the bee's alimentary canal and then pass into the intestines where it is stored in the rectum until it is excreted. From time to time people get alarmed about a phenomenon referred to as "yellow rain" (Newman 1984). Yellow rain is nothing more than bee feces. When large numbers of bees forage on nectar sources containing high quantities of pollen, the rapid removal of those pollen grains from their honey stomachs quickly fills their rectums. The result can be rapid defecation by those swarms of bees as they return to the hive. If the flight path of the bees happens to be over urban areas, their feces, or "yellow rain", may leave hundreds of tiny yellow spots on cars, sidewalks, or buildings.

The development and use of pollen coefficients in melissopalynology

One of the primary goals of Todd and Vansell (1942) was to determine how effectively honeybees could remove pollen from their honey stomachs, how long that pollen removal process took, and if all pollen types were removed equally well by the filtering process of a bee's honey stopper. In one experiment they fed a laboratory beehive diluted unifloral star thistle (Centaurea) honey that had been produced by bees foraging in the wild. The star thistle pollen concentration in the honey-water mixture was measured as being 5,200 pollen grains per cc. Later, the sealed honeycomb cells produced by caged honeybees feeding on this honey-water source were removed and examined. Todd and Vansell found that instead of an average of 5,200 star thistle pollen grains per cc, the produced honey contained an average of only 1,200 pollen grains per cc. Even though Todd and Vansell had expected the pollen concentration of the newly produced honey to remain constant, they found that it did not. In another experiment Todd and Vansell mixed three grams of pure pollen (the pollen type they used is not mentioned) with 100 cc of water-diluted syrup. When measured, they found the pollen concentration of the diluted syrup solution was 750,000 pollen grains per cc. After allowing honeybees to feed on that mixture as frequently as they wanted, the researchers removed honeycomb cells made from that syrup and discovered that the pollen concentrations of the new honey were only 25,000 pollen grains per cc. In other words, the honeybees drank a diluted syrup solution containing a pollen concentration of 750,000 pollen grains per cc, then, using their internal honey stomach filtration system those bees removed most of the pollen before emptying their honey stomachs into new comb cells. The result was a newly produced honey containing only 1/30th of the original pollen concentration of the diluted syrup. Although these researchers' goals were not to develop pollen coefficient tables, their pioneering effort led others to use their ideas and experimental data to compile lists of plants that are over or under represented by their pollen in honey samples (Maurizio 1949, 1955, 1958; Berner 1952; Pritsch 1957; Deans 1957; Demianowicz 1961, 1964; and Sawyer 1988) and propose pollen coefficient tables for various types of nectar-producing plants.

Demianowicz (1961, 1964) is one of the early melissopalynologists who worked tirelessly for many (13) years trying to solve the problem of accurate unifloral honey classification based on pollen contents. After examining many honey samples Demianowicz realized that the relative pollen count in honey did not always reflect the primary floral and nectar sources. Demianowicz's summarized data appear in her 1964 publication where she attempts to identify the pollen characteristics of 45 different types of unifloral honey that are common to various regions of Europe. For each unifloral type she used caged bees in small hives of only 300-400 workers and one queen. The bees in each caged hive were allowed to feed on the flowers of only one plant species. Thus, the honey each hive produced was considered to be a valid representation of the expected absolute pollen concentration (APC) for the flower type being tested. Based on that research, she developed 18 different categories of plants based on whether their APC values in honey are under or over represented. For each category she assigned an "average number" that she called her "pollen coefficient classes". She believed that the newly established pollen coefficient values could be used as a guide for determining the true unifloral nature of a honey sample, regardless of the data represented by the relative pollen concentrations.

In Demianowicz's table of values she says that the expected APC of pollen types in a "class 1 unifloral type" should not be expected to be higher than 740 pollen grains per 10 g of honey. Her key example of an under represented type in class 1 is Asclepias (milkweed) which has an assigned pollen coefficient value of 32. Each additional class is represented by APC values that are up to twice as high as the previous category. Thus, in her class 2 of unifloral honey types she says that genera in this group should contain between 750-1,500 pollen grains per 10 g of honey. Plant examples in this second category include Robinia pseudoacacia (white acacia, locust), Cucumis (cucumber), and Chamaenerio (Epilobium) (fireweed). The last of her coefficient categories is class 18, which is characterized by prolific pollen types, such as Myosotis (forget-me-not ), which produce unifloral honeys containing between 98,304,001 to nearly 200 million pollen grains per 10 g.

Many melissopalynologists have worked on the problem of trying to discover how to use pollen contents to classify various types of unifloral honey even though experimental data (Todd and Vansell 1942) show that plants produce different amounts of pollen and that bees will remove vast amounts of pollen from collected nectar on their flights back to the hive. Nevertheless, a number of scientists have produced tables and charts noting what they believe should be the "expected" percentages of relative pollen in unifloral types. Moar (1984) offers one of the clearer discussions about the process that he used to establish pollen coefficient tables for various types of unifloral honey in New Zealand. Moar points out that since 45% of a single pollen type is the universal "minimal" amount needed for a honey to be classified as unifloral, one must determine what value must be used to correct an under represented unifloral pollen type to the 45% level. Next, he points out that he and others believe that the relative pollen values for white clover (Trifolium repens) are the best standard upon which all other types in honey samples must be judged. Therefore, it is chosen to be the standard pollen type for determining the coefficient values of all other pollen types. Moar notes that the expected absolute pollen concentration of Trifolium repens pollen should be approximately 23,116 grains per 10 g of honey. However, Moar fails to explain exactly how he determined this figure to be the APC value for T. repens. For example, Demianowicz's (1964) research with caged bees revealed that 18,000 should be considered the APC for a unifloral honey of T. repens. Perhaps the differences between these two researchers derive from their different methods of calculating the APC value for T. repens. One of the most widely used calculation methods is to add a known number of "tracer spores" to 10 g of honey and then determine the ratio of tracer spores to pollen in the honey. Once this ratio is known for a sample, then the APC for any pollen taxa in the sample can also be determined. Demianowicz's calculation method is different. For each of her honey samples she began with a known amount of honey that was diluted with a known amount of water. From that mixture she extracted a small portion and counted the pollen. From that count she then predicted what the total APC for each pollen type should be per 10 g of honey.

In his 1985 article on the honey of New Zealand, Moar explains how he calculated the minimum percentage of unifloral honey types. For his example of an under represented type he used thyme (Thymus) pollen. He says his first task was to find examples of honey that was produced by hives located close to fields of blooming thyme. If the honey from those hives had the color and taste of thyme honey, then they were assumed to be unifloral examples of thyme honey. Moar found four samples of honey that fit these criteria. When he conducted a pollen study of these, he found that the relative pollen percentage averaged 42%, even though the totals varied slightly in each sample. He also averaged the pollen concentration values in the four thyme samples and found that the APC of thyme pollen was 5,415 pollen grains per 10 g of honey. Because the relative pollen percentage of thyme was less than the minimum of 45% needed for a unifloral classification, Moar multiplies the thyme's APC by 0.45 and then divides by the average relative pollen frequency, (0.42) of thyme in the four samples. That math calculation produced the number 5,801, which Moar points out should be considered as the appropriate "corrected" APC for thyme pollen at the minimum unifloral level of 45%. Nevertheless, because thyme pollen is an under represented taxon in honey (i.e., any taxon with an APC lower than the APC of white clover), a further calculation was needed to determine the minimal percentage of thyme pollen in the relative pollen count of a honey sample before the sample could be classified as being unifloral thyme. Moar then notes that because the APC of white clover is considered the standard, the ratio of thyme's APC of 5,801 to the APC (23,116) of white clover must be determined. This is calculated by dividing the APC for thyme (5,801) by the combined APC of thyme and white clover (5,801 + 23,116). That number (0.2), two-tenths, is then multiplied by 100 to convert it to a percentage (20%). As Moar explains, the use of these calculations reveal that for New Zealand, a relative pollen percentage of only 20% thyme pollen in a honey sample would qualify that honey as being a unifloral thyme honey.

One of the foremost melissopalynologists in the United Kingdom (U.K) is Rex Sawyer who began his lifelong interest in pollen studies during the 1930s after meeting Harry Godwin, the British pioneer of pollen analysis. Later, Sawyer began a study of beekeeping and helped Deans (1957) compile a detailed pollen study of honey types produced in the U.K. After years of melissopalynology research Sawyer published several books on pollen and honey (Sawyer 1981, 1988). In one of them (Sawyer 1988) he includes a chapter on pollen coefficients and lists a table of numerical pollen coefficient (PC) values that he developed and believes can be applied to the relative pollen percentages of various pollen types in honey samples. The table is impressive, yet Sawyer fails to explain exactly where or how he arrived at the precise PC values that he published. His only reference to this is a note saying that his values are based on APC data from his own research and from various other studies including the research of: Todd and Vansell (1942), Demianowicz (1961, 1964), Maurizio (1949, 1955, 1958), Berner (1952), and Pritsch (1957).

The primary difference between the PC values used by Sawyer and those published by other melissopalynologists is that Sawyer's values are not expressed as the expected APC per gram of honey for each taxon whereas others all use APC values per 10 g of honey. Nevertheless, by using Sawyer's formula and applying his PC numbers to the relative pollen percentages of a honey sample one can determine the "actual" floral identity and characteristics of almost any honey sample. Sawyer notes that once the expected relative pollen percentages for key pollen taxa are determined for a honey sample, it is possible to use his formula and PC values to calculate the "corrected percentages" for nearly all pollen types found in both unifloral and mixed floral honeys. He also notes that since he did not calculate the PC values for all known pollen types in honey, he recommends using a PC value of 50 for all unidentified pollen types and pollen taxa that appear in sporadic or low frequencies.

The result of Sawyer's research led to the establishment and use of his table of pollen coefficient numbers for honey types in the UK as well as elsewhere. As Sawyer and others have argued, through the use of pollen coefficient tables analysts can confirm unifloral honey samples produced from plants with low pollen yields or plants which produce pollen types that seem to be most quickly removed by the filtering actions of a honeybee's honey stopper during her return flight to the hive. As noted in Sawyer's list, some of the weakest-represented pollen types are aided by the use of pollen coefficient values. These would include fireweed (Epilobium), basswood (Tilia), alfalfa (Medicago), sourwood (Oxydendron), orange blossom (Citrus), buckwheat (Eriogonum), and locust (Robinia pseudoacacia). When pollen coefficient tables are applied to correct the normally low relative pollen percentages, of up to 20%, for these types in honey samples, the result becomes a validation that the honey should indeed be classified as a unifloral product of those plants.

An example of how pollen coefficient tables developed by Sawyer (1988) can be used is seen in Table 1. The data in Table 1represent the analysis of a typical fireweed honey sample from central Alaska. Note that based on the relative percentages of pollen, this sample would initially be classified as a "uniflora" canola honey. Note also that the relative percentage of fireweed pollen in this sample is only 6.3%. When these relative pollen percentages are adjusted using Sawyer's pollen coefficient values, it suggests that the actual nectar sources used to produce this honey were really 95% from fireweed flowers, and only 2% from rapeseed (canola) flowers. These pollen data suggest that in spite of the low relative pollen percentage of fireweed, the dominant nectar resource that produced this honey came from blooming fireweed flowers. Note also in this example (Table 1) that the 28.3% relative pollen percentage of clover suggests that actually less than 2% of the nectar source of this honey came from clover.

TABLE 1
Pollen analysis of a honey sample produced in central Alaska.*
Pollen Type	Relative Pollen %	Coefficient Value	Relative Quantity	Adjusted %
Apiaceae	00.6	50.0	00.012	00.5
Brassica	62.8	150.0	00.419	01.9
Epilobium	06.3	0.3	21.000	95.9
Melilotus	28.3	75.0	00.377	01.7
Taraxacum	00.6	10.0	00.060	00.27
Unknown	01.4	50.0	00.028	00.128
Total	100.0%		21.896	100%
* To use pollen coefficient tables to determine the actual or expected nectar composition of each plant taxon in a honey sample, the relative pollen spectrum must first be calculated. Next, the relative percentage of each pollen type must be divided by its pollen coefficient value (Sawyer 1988). The resulting value for each pollen type is what Sawyer calls the taxon's "relative quantity". Finally, the percentage of each pollen type's relative quantity is used to determine what percentage of the honey is derived from the nectar represented by each taxon.

One of the main problems of using pollen coefficient tables is that not all melissopalynologists agree on precisely what values should be assigned to each plant and pollen type used by bees to produce honey. For example, if we examine the same data listed in Table 1, but instead use those relative pollen percentages with the pollen coefficient values developed by Demianowicz (1964), we find that the adjusted percentages of nectar sources for each taxon is different from those of Sawyer's (Table 2).

TABLE 2
In the example below we have used the same relative pollen percentages shown for taxa in Table 1. As Sawyer (1988) recommends, we have used the baseline PC value of Trifolium repens for unknown pollen types or pollen types for which no PC values were determined by Demianowicz (1964). In the formula below, "A" represents the adjusted percentage for each pollen taxon.
Aa = (a/PCa)/(a/PCa + b/PCb + c/PCc + d/PCd + ........ n/PCn)
Aa = .000033/(.000033 + .000872 + .005600 + .000393 + .000033 + .000077)
Aa = .000033/.007008
Aa = .005 = .47%
Taxon	Relative %	Demianowicz's adjusted %	Sawyer's adjusted %
a = Apiaceae	00.6	00.47	00.50
b = Brassica	62.8	12.44	01.90
c = Epilobium	06.3	79.90	95.90
d = Melilotus	28.3	05.60	01.70
e = Taraxacum	00.6	01.09	00.27
f = unknown	01.4	00.47	00.12
Total	100%

When examining the "adjusted percentages" calculated using Sawyer's data and those derived from Demianowicz's data we can see minor differences; nevertheless, both sets of data offer the same conclusion. In terms of over or under representation, both sets of data indicate the same general conclusions for each pollen taxon. For example, using Sawyer's PC values it appears that over 95% of the actual nectar source came from fireweed flowers, but when using Demianowicz's values it appears that only 80% of the nectar source came from fireweed flowers. There is a difference of 15% between the two calculations, yet both sets of data emphasize that fireweed is a pollen type that is highly under represented in honey samples. Similar conclusions can be reached for each of the other pollen types as well.

It is difficult to know which of these PC data sets (Sawyer vs. Demianowicz) is more nearly accurate. Demianowicz's PC values are based on more than 13 years of research in which she used caged bees and forced each hive to feed only one type of flower. From those data she constructed her APC and PC values. We are not sure which data Sawyer used to construct his PC values. However, based on her published reports we do know the method Demianowicz (1961, 1964) used for calculating pollen concentration values in her study and we also know that her method is subject to more potential calculation errors than most of the currently-used pollen concentration techniques. Demianowicz's method for determining pollen concentration values relied on collecting small amounts of honey (often less than 5 grams) and then diluting her sample with water. From the diluted solution she then placed one drop on a microscope slide and counted a portion of the pollen on the slide. Using that information she then projected what the expected APC values for each pollen taxon should be in 10 g of honey. By contrast, today most melissopalynologists calculate pollen concentration values by comparing the ratio of pollen grains found in 10 g of honey against the ratio of a know number of tracer spores that are added to the honey before counting begins. Depending on the number of tracer spores added and the number of pollen grains counted, this system is considered to provide more nearly accurate data than the system used by Demianowicz.

Other melissopalynologists (D'Albore 1998; van de Ham et al. 1999) have not proposed pollen coefficient tables but do list a wide variety of plant types, which they place into various categories depending on the "expected" pollen, yield of those plants in various types of honey. For example, van de Ham et al. (1999) mentions that honey types can be considered unifloral examples if they contain a minimum of: 1) Borago [borage] 10%; 2) Robinia [white acacia, locust], Tilia [linden, basswood], and Carduus [distel, thistle] 20%; and 3) Crambe [krambe, sea kale] and Calluna [heather] 30%. They also state that for some plants the APC of their pollen is so prolific in honey that one should not consider those honey types as unifloral unless they contain significantly more than the normally required minimum of 45%. Some of those over-represented pollen types they mention include: 1) Salix [willow] 70%; and 2) Phacelia [bluebells], Myosotis [forget-me-not] and Castanea [chestnut, chinkapin] each at 90%.

D'Albore (1998) does not give lists of percentage levels that he believes should be used as a guide for determining unifloral honey in the Mediterranean region based on the under or over-representation of pollen taxa. Nevertheless, he offers some practical advice about a number of plant taxa and how the pollen from those taxa is likely to be represented in honey samples. He notes that most of the plant group that produces large pollen grains (>40 m) will be significantly under represented in honey produced from those nectar sources. The two reasons he gives for this are: 1) most plants that produce large pollen grains generally do not produce large quantities of nectar, and 2) honeybees are much more efficient at filtering out large pollen grains than small ones from the nectar in their honey stomach during their return flights to the hive. D'Albore adds that the opposite is true for tiny pollen grains, which are most often over represented in honey. Tiny pollen grains from species including Echium [gran canaria], Eucalyptus [gum], Amorpha [indigo], Castanea [chestnut, chinkapin], and Tamarix [salt cedar], he notes, are often produced in larger numbers and those tiny grains are only partially filtered out of the honey stomach of honeybees returning to the hive. In his book he also lists a large number of plants in the Mediterranean region that produce pollen, which tends to be either over or under represented in honey samples for a variety of reasons. The reasons he lists include: 1) plants that normally produce small amounts of pollen (i.e., Citrus, Robinia, Salvia); 2) plants that are monoecious and thus only one half of the flowers produce pollen (i.e., Citrullus [watermelon], Cucumis [cucumber], Cucurbita [pumpkin or gourd], Bryonia [bryony]); 3) plants that have flowers that are morphologically unfavorable for pollen collection by honeybees (i.e., Asphodelus [affofill], Epilobium [fireweed], Abutilon [mallow], Datura [datura], Digitalis [foxglove]); and 4) plants that present special pollen and nectar gathering problems for honeybees or have plants that are difficult for honeybees to enter (i.e., Agrostemma [corn cockle], Cestrum [cestrum], Nicotiana [tobacco], Medicago [alfalfa]).

Todd and Vansell's (1942) research revealed other important variables that will determine the amounts and types of pollen recovered from honey samples. In one experiment they starved a group of honeybees and then allowed them to drink freely from solutions of sugar syrup mixed with various amounts of pollen. After feeding a control group of bees was immediately trapped and dissected. The content of their honey stomach was removed and examined for pollen. Bees that fed on the syrup-pollen mixture had an average of 248,666 pollen grains per cc of fluid in their honey stomachs. Forty-eight other bees that fed on the same syrup-pollen solutions were allowed to fly around freely for 15 minutes after feeding before being caught and dissected. The same procedure was used to determine the pollen concentration values of the fluid in their honey stomachs. As noted in those tests, almost one-half of the honeybees were able to remove and excrete more than 90% of the pollen they had consumed when feeding on the syrup-pollen solutions. The other bees that were tested in that experiment also removed much of the pollen. These data suggest that even though all 48 of the bees in this experiment collected nectar containing the same amount of pollen, some of them were much more effective at removing pollen with their honey stoppers than were others from the same hive. This variable makes the reliability of using a standard set of numbers for pollen coefficient tables difficult. For any given honey sample the accurate calculation and use of corrective PC values are nearly impossible unless the melissopalynologist knows how many honeybees collected nectar for each source, how many of those honeybees removed what percentage of the pollen from the nectar they collected, and how long each honeybee took on her return flight to the hive after filling her honey stomach full of nectar. Since these types of nectar gathering data can only be estimated for any hive or any type of honey a hive produces, the use of PC tables could vary the results greatly from one sample to the next even if all examined honey samples were produced from the same nectar sources, but for some samples those sources were located at different distances from the hive.

Todd and Vansell repeated this same experiment with different pollen concentrations in a sugar syrup solution that they fed to honeybees. In all their tests they found that the amount of pollen still present in the honey stomachs of bees allowed to fly freely for 15 minutes after feeding was drastically reduced. Although the results varied from bee to bee, Todd and Vansell reported that many of the bees had an ability to remove at least 90% of the pollen from the fluid in their honey stomach during a 15-minute interval after feeding on nectar containing pollen.

Another contribution of the Todd and Vansell (1942) study was the development of a table revealing how many pollen grains occur naturally in the nectars of certain plants. Because flower nectar sources are usually close to the dehiscing anthers of those same flowers, some of the anther pollen falls into the nectar that is later gathered by honeybees. Todd and Vansell carefully collected the nectar from more than 2,600 samples representing 73 different plant species that grow in California. Some of those samples were collected from the honey stomachs of bees that were captured and dissected immediately after they fed on the nectar of a specific plant. Other samples were carefully collected by hand from the nectar of actual flowers. After all the samples were examined and the pollen concentrations of each nectar source were averaged, the authors produced a list of some of the major California nectar sources and the amount of pollen that one might expect to find in the nectar of those plants. That list is important because it offers a perspective as to which nectar types are known to contain vast amounts of pollen and which nectar sources do not. For example, Todd and Vansell report that they captured and dissected 30 honeybees immediately after each had finished feeding on orange blossoms (Citrus sinensis) and they also collected a set of 32 bees immediately after each had fed on cotton flowers (Gossypium hirsutum). To their amazement, they could not find one single pollen grain in the honey stomachs of any of those 62 honeybees. At the other extreme, they found an average of 7,100 pollen grains per cc of fluid in the honey stomachs of 38 bees captured immediately after each had completed feeding on the nectar of rabbit brush (Chrysothamnus nauseosus).

In another experiment Todd and Vansell examined what happens to pollen between the time it is collected as part of the nectar from a flower until it becomes honey sealed in the comb chamber of a hive. They deprived the honeybees in a hive of stored honey but allowed them to feed from trays placed in the hives that were filled with a solution of sugar syrup mixed with pollen. They repeated the same experiment in a different hive using a tray of diluted star-thistle honey placed in the hive. Measurements of the pollen in the feeding trays revealed a pollen concentration value of 750,000 pollen grains/cc for the tray of syrup mixed with pollen and 5,200 pollen grains/cc for the feeding tray of diluted star-thistle honey. During both experiments the bees were not allowed to feed on any other sources of food. Honey produced in the sealed comb cells made from these two solutions revealed a pollen concentration of 25,300 pollen grains/cc for the honey made from the syrup-pollen solution and 1,200 pollen grains/cc for the honey made from diluted star-thistle honey. Todd and Vansell concluded from these experiments that only 3.1% of the pollen placed in the syrup-pollen feeding trays actually appeared in the honey made from that source. For the honey made from the feeding tray of diluted star-thistle honey, only 23% of the pollen from the tray appeared in the new combs of honey. In both experiments the feeding trays were placed in the hives; therefore, the probable time between each bee's feeding time and when she regurgitated the contents of her honey stomach into comb cells was probably minimal. In spite of that suspected short period, it appears that most of the honeybees in both hives were able to remove vast amount of pollen from the fluids they ingested before those fluids were used to make honey.

Following up on this experiment Todd and Vansell tested actual nectar sources from the flowers of important bee-foraging plants and found an average of 2,500 pollen grains per cc in purple sage (Salvia leucophylla), an average of 200 pollen grains per cc in fireweed (Epilobium angustifolium), an average of 2,000 pollen grains per cc in avocado (Persea americana), and an average of 41,000 pollen grains per cc in white sweet clover (Melilotus alba). What those data reveal is that the pollen of some plants will always be either under or over represented in honey even if 90% (i.e., average efficiency of pollen removal by most bees based on experimental data by Todd and Vansell) of the pollen is removed from the honey stomach of a honeybee returning to the hive. What these data suggest is that one cc of nectar from sweet clover should be over 200 times better represented by its pollen in a honey sample than will an equivalent cc of nectar from fireweed plants.

In recent years the intended purpose of producing and using data that list over and under-represented pollen types in honey (D'Albore 1998; van de Ham et al. 1999), or lists that detail the actual pollen coefficient number for each taxon (Sawyer 1988; Demianowicz 1961, 1964), is to allow individuals to use the relative pollen percentages from a honey sample to re-calculate the probable (actual?) percentages of each nectar source used to produce that honey sample. The main advantage of using these data is to verify unifloral honey types that command premium prices on the world market even though the relative pollen counts from such honey samples probably do not contain the internationally accepted minimum pollen percentage of 45%.

Tables of under and over represented pollen types and pollen coefficient tables also help to explain why certain popular bee foraging plants are routinely represented by minimal amounts of pollen in honey that may be dominated by pollen from other types of over represented bee-foraging plants such as Melilotus and Brassica. Research on developing better pollen coefficient tables continues, but some melissopalynologists do not believe that these types of "correction" tables will ever become universally accepted. Arguments against the adoption of a standard set of pollen coefficient tables focus on several factors, which are rarely known for any given honey sample. For example, as early as the 1920s scientists knew that the longer nectar remains in a honeybee's honey stomach, the greater is the potential for that honeybee to remove most or all of the pollen in that nectar, regardless of the pollen type (Whitcomb and Wilson 1929). Therefore, knowing the time period between when a honeybee begins to forage and when she returns to the hive becomes critical because it will influence the amount of pollen that remains in the nectar of her honey stomach. That information is important because it determines the amount of potential pollen from each floral source that can be included in the honey produced from each nectar source. The second variable focuses on the size and shape of the pollen grains being collected along with the nectar from a floral source. Experimental data reveal that bees are much more efficient at removing large pollen grains from the nectar in their honey stomachs than they are for smaller pollen grains (Demianowicz 1961, 1964; D'Albore 1998). The third variable centers on determining precisely what the corrective PC value for each pollen type should be. The published data presented by various melissopalynologists state different APC or PC values for the same or similar plants. Often these values are somewhat similar, but depending on which APC or PC values a researcher selects to use, the actual percentage of a single pollen type needed for a honey to gain a unifloral classification will vary.

Summary

The use of corrective pollen values is important for beekeepers, honey distributors, and for the customer who buys honey for his or her own use. Unfortunately, much of the existing research in this area of honey studies suffers from one or more major flaws. Some previous researcher have not provided critical information on how they gathered, processed, or counted the pollen in the honey samples they used to establish their APC or PC tables. Others offer full explanations of their research and thus reveal the flaws of their methodology. Even the research conducted by Demianowicz (1961, 1964), which is among the best studies of APC and PC values yet completed because she used caged bees that ate a restricted diet, there are flaws in her technique which could have altered her results.

What is needed most in the field of melissopalynology is a new series of tests to determine the precise PC values that can be used with certainty for validating the floral sources of premium types of honey. This type of research, however, will need to be conducted under controlled conditions that will satisfy skeptics and produce PC data that will be accepted by melissopalynologists.

For each floral type a separate experiment will need to be conducted. First, an isolated hive of bees will need to be caged to prevent outside contamination, similar to the technique reported by (Demianowicz 1964). The caged bees should then be allowed to feed freely on the flowers of only one type of plant until the hive produces a measurable amount of honey from that single floral source. During the experiment selected numbers of bees should be trapped immediately after they have fed on nectar and the contents of their honey stomach must be examined to determine the APC in the nectar. At various times during the experiment the opening for bees returning to the hive should be sealed for short periods of time ranging from 5-15 minutes. Before allowing the bees to re-enter the hive, some of the bees should be captured and the contents of their honey stomachs should be examined to determine how effective those bees have been at removing pollen from the nectar they are collecting. Finally, once honey has been produced from the experimental feeding process, several honey samples must be collected, processed, and their pollen contents counted.

Processing of the collected honey from these caged experiments must include either a filtration process similar to the one described by Lutier and Vaissiere (1993), or it must use an alcohol-dilution technique similar to the one first described by Jones and Bryant (1996). Later experimental tests conducted by Jones and Bryant (1998) confirmed that both the filtration and alcohol-dilution techniques are comparable in the amounts of pollen they recover from honey samples. Their tests revealed that both the filtration and alcohol processing methods increased pollen recovery from honey samples by an average of more than 200% over the various types of water-dilution processing methods that are currently in use by melissopalynologists. Finally, a large quantity of tracer spores must be added to each honey sample before it is processed. The original ratio of tracer spores to pollen should be nearly equal in each honey sample to ensure that the construction of APC tables for pollen types are as accurate as possible. Finally, when counting the recovered pollen from each honey sample, the ratio of tracer spores to pollen must be based on high pollen counts in excess of 1,000 pollen grains per sample.

This type of proposed research will be expensive and time consuming. However, if these research efforts are completed successfully, the resulting data can be confidently used to construct APC values that should be accepted by even the most ardent skeptics. In addition to the APC values, these same honey samples can serve as unique opportunities for chemical testing to determine sugar types and the ideal ranges for sugar isotopic levels.

Bibliogaphy

ALLEN, M. Y.
1928a Armchair thoughts. Bee World 9: 57.
1928b Pollen grains. Bee World 9: 66.
1928c Pollen grains II. Bee World 9: 103-105.
1928d Pollen grains III. Bee World 9: 148-151.
1929 Pollen grains IV. Bee World 10: 114-118.

ALFONSUS, E. O.
1933 Zum Pollenverbrauch des Bienenvolkes. Archiv für Bienenkunde 14: 220-223.

ARMBRUSTER, L.
1921 Vergleichende Eichungsversuche auf Bienen und Wespen. Archiv für Bienenkunde 3: 219-230.

BERNER, U.
1952 Die ausvertung der pollenanalyse. Archiv für Bienenkunde 29:33-38.

BETTS, A. D.
1923 The identification of pollen sources. Bee World 5: 43-45.
1925 Notes on pollen identification. Bee World 7: 90.

CRANE, E.
1975 Honey, a Comprehensive Survey. Russak & Co., Inc., New York.

D' ALBORE, G. R.
1998 Mediterranean Melissopalynology. Universita Degli Studi di Perugia, Instituto di Entomologia agrania Publication. Perugia, Italy. 466 pages,

DEANS, A. S. C.
1957 Survey of British honey sources. International Bee Research Association Circular, London.

DEMIANOWICZ, Z.
1961 Pollenkoeffizienten als Grundlage der quantitativen Pollenanalyse des Honigs. Pszczelnicze Zeszyty Naukowe 5(2): 95-105.
1964 Characteristik der Einartenhonige. Annales de l'Abeille 7: 273-288.

DIETZ, A.
1975 Nutrition of the adult honeybee. In: The Hive and the Honey Bee, (Dadant, C., and Dadant, C. P.,eds.), Dadant & Sons, Carthage, Illinois, p. 125-126.

ECKERT, J. E.
1942 The pollen required by a colony of honeybees. Journal of Economic Entomology 35: 309-311.

EUROPEAN ECONOMIC UNION
2001 Web site http://europa.eu.int/eur-lex/en/lif/dat/1974/en_374L0409.html

GARY, N. E.
1975 Activities and behavior of honeybees. In: The Hive and the Honey Bee, (Dadant, C., and Dadant, C.P.,eds.), Dadant & Sons, Carthage, Illinois, p. 185-264.

HAYDAK, M. H.
1935 Brood rearing by honeybees confined to pure carbohydrate diet. Journal of Economic Entomology 29: 657-660.

JONES, G. and BRYANT, V.
1993 Melissopalynology. In: Palynology: Principles and Applications, (J. Jansonius and D. C. McGregor eds.), pp 933-938. American Association of Stratigraphic Palynologists Foundation, Dallas, Texas.
1996 Techniques in Melissopalynology. In: New Developments in Palynomorph Sampling, Extraction, and Analysis (V. Bryant & J. Wrenn, eds.), pp 107-114, AASP Contribution Series # 33. American Association of Stratigraphic Palynologists, Dallas.

LIEUX, M. H.
1969 A palynological investigation of Louisiana honeys. Unpubl. doctoral dissertation, Department of Botany and Plant Pathology, Louisiana State University, U.S.A., 113 p.
1975 Dominant pollen types recovered from commercial Louisiana honeys. Economic Botany 29: 78-96.
1977 Secondary pollen types characteristic of Louisiana honeys. Economic Botany 31: 111-119.
1978 Minor honeybee plants of Louisiana indicated by pollen analysis. Economic Botany 32: 418-432.

LUTIER, P. M., and VAISSIERE, B. E.
1993 An improved method for pollen analysis of honey. Review of Palaeobotany and Palynology 78: 129-144.

LOUVEAUX, J., MAURIZIO, A., and VORWOHL, G.
1970 Methods of melissopalynology. Bee World 51: 125-131.

MAURIZIO, A.
1949 Pollenanalytische Untersuchungen an Honig und Pollenhöschen. Beihefte zur Schweizerische Bienen-Zeitung 2: 320-412.
1951 Pollen analysis of honey. Bee World 32: 1-5.
1955 Beiträge zur quantitativen Pollenanalyse des Honigs. 2. Absoluter Gehalt pflanzlicher Bestandteile in Tilia- und Labiaten-Honige. Zeitschrift für Bienenforschung 3(2): 32-39.
1958 Beiträge zur quantitativen Pollenanalyse des Honigs. 3. Absoluter Gehalt pflanzlicher Bestandteile in Esparsette, Luzerne, Orangen und Rapshonige. Annales de l'Abeille 11: 93-106.

MOAR, N. T.
1985 Pollen analysis of New Zealand honey. New Zealand Journal of Agricultural Research 28: 39-70.

NEWMAN, C.
1984 Pollen: breath of life and sneezes. National Geographic Magazine (October):490-521.

OERTEL, E.
1939 Honey and pollen plants of the United States. U. S. Department of Agriculture Circular 554, Washington, D. C.

PAMMEL, L. H., and KING, C. M. (eds.)
1930 Honey Plants of Iowa. Iowa Geological Survey Bulletin 7. State of Iowa, Des Moines, Iowa.

PARKER, R. L.
1923 Some pollen gathered by bees. American Bee Journal 63: 16-19.

PELLETT, F. C.
1930 America Honey Plants, Together with Those Which are of Special Value to the Beekeeper as Sources of Pollen. (3rd ed.) American Bee Journal, Hamilton, Illinois.

PFISTER, R.
1895 Versuch einer Mikroskopie des honigs. Forschber. Lebensmitt. u. ihre Bez. z. Hygiene, Forens. Chem., Pharmakogn. (2)(1/2): 1-9; 29-35.

PRITSCH, G.
1957 Zum problem der mikroskopischen pollenanalyse des bienenhoring. Wiss. Zs. Humboldt-univers. Berlin 6:197-204.

SAWYER, R.
1988 Honey Identification. Cardiff Academic Press, Cardiff, Wales, UK.

SEELEY, T. D.
1985 Honeybee Ecology. Princeton University Press, Princeton, New Jersey.

SNODGRASS, R.E. and ERICKSON, E. H.
1992 The anatomy of the honey bee. In: Dadant and Sons (eds.) The Hive and the Honey Bee, Dadant & Sons, Hamilton, Carthage, Illinois, p. 103-169.

TODD, F. E., and VANSELL, G. H.
1942 Pollen grains in nectar and honey. The Journal of Economic Entomology 35: 728-731.

VAN DER HAM, R.W.J.M., KAAS, J.P., KERKVLIET, J.D. and NEVE, A.
1999 Pollenanalyse. Stichting landelijk Proefbedrijf voor Insektenbestuiving en Bijenhouderij Ambrosiushoeve. Hilvarenbeek, Netherlands. 156 p.

WHITCOMB, W. JR., and WILSON, H.F.
1929 Mechanics of digestion of pollen by the adult honeybee and the relation of undigested parts to dysentery of bees. Wisconsin Agricultural Experiment Station Research Bulletin 92.

WINSTON, M. L.
1987 The Biology of the Honey Bee. Harvard University Press, London.

WODEHOUSE, R. P.
1935 Pollen Grains. Hafner, New York.

YOUNG, W. J.
1908 A microscopical study of honey pollen. United States Bureau of Chemistry, Bulletin 110, Washington, D.C. 93p.

ZANDER, E.
1935 I. Pollengestaltung und Herkunftsbestimmung bei Blütenhonig. Reichsfachgruppe Imker, Berlin.
1937 II. Pollengestaltung und Herkunftsbestimmung bei Blütenhonig. Liedloff, Loth, & Michaelis, Leipzig.
1941 III. Pollengestaltung und Herkunftsbestimmung bei Blütenhonig. Liedloff, Loth, & Michaelis, Leipzig.
1949 IV. Studien zur Herkunftsbestimmung bei Waldhonigen. Ehrenwirth, Munich.
1951 V. Letzte Nachträge zur Pollengestaltung und Herkunftsbestimmung bei Blütenhonig. Liedloff, Loth, & Michaelis, Leipzig.

---

Author contact: Vaughn M. Bryant, Jr, Palynology Laboratory, Texas A&M University, College Station, Texas 77843-4352, USA

From CAP Newsletter 24(1):10-24, 2001.