Intrahousehold allocation of resources in larger and smaller Mexican households

Intrahousehold allocation of resources in larger and smaller Mexican households

Sm. Sci. Med. Vol. 36, No. 3, pp. 305-310, 1993 Printed in Great Britain.All rightsreserved 027%9536/93 $6.00 + 0.00 Copyright0 1993PergamonPressLtd...

701KB Sizes 2 Downloads 70 Views

Sm. Sci. Med. Vol. 36, No. 3, pp. 305-310, 1993 Printed in Great Britain.All rightsreserved

027%9536/93 $6.00 + 0.00

Copyright0 1993PergamonPressLtd

INTRAHOUSEHOLD ALLOCATION OF RESOURCES IN LARGER AND SMALLER MEXICAN HOUSEHOLDS ROBERTAD. BAER’.*and LORENAMADRIGAL’ ‘Department of Anthropology, University of South Florida, Tampa, FL 33620, U.S.A. and *Department of Nutrition, Centro de Investigation en Alimentacion y Desarrollo, Hermosillo, Sonora, Mexico Abstract-This paper focuses on aspects of intrahousehold allocation of resources. It suggests that there is a first step involved in understanding household decisions as to allocation or misallocation of food within the household, which is that of understanding intrahousehold allocation of income. In this study, carried out in northwestern Mexico, what has been reported in the literature as examples of unequal access to food, particularly in larger households, may better be considered examples of lack of access on the part of purchasers of household food to all of the income which comes into the household. Key words-intrahousehold

allocation of resources, food consumption, Mexico

INTRODUCTION This paper brings together two aspects of intrahousehold dynamics, intrahousehold allocation of income and intrahousehold distribution of food, and suggests that understanding the former is an important prerequisite not only to understanding the latter, but also to evaluating if there is misallocation of food within the household. It is suggested that differences in the amount of income available to those purchasing food for these households may be a factor of critical, but often unexamined importance in understanding differences in patterns of food consumption and nutritional status in households of different sizes. In understanding issues of food consumption, increasing attention is being paid to intrahousehold dynamics. It is recognized that household behavior is a critical factor in dietary and nutritional outcomes [l]. One type of research in this area has focused on studies of the intrahousehold distribution of food. While there are still many unresolved methodological problems to be overcome in these studies [2], a correlation between malnutrition and household size has been observed [3-51. However, in the latter study, the authors assumed that all the households were of a similar income level, because the study was carried out in a low income area of a Colombian town. While this may or may not have been the case at the level of the household, it is most unlikely that this was the situation with respect to per capita income. Thus, the observed correlation between large family size and malnutrition may in reality by a reflection of per capita income differences among the households. When per capita income is controlled, the problem may still not be one of “competition for food among members of large households which results in a tendency toward malnutrition among its youngest segment” [4, p. 2261. It has been suggested that larger households may allow decreased costs per person for

the same amounts of food because of the possibility of purchasing food in bulk. Larger households may also have less wasted food than do smaller households because of the presence of more people to consume the food before it spoils [6]. Dietary intake data from Bangladesh revealed this pattern of a positive association between nutrient intake and household size [6]. Discrimination in favor of adults (vs children) in intrahousehold allocation of food was not found, when allowance was made for differential needs [7]. However, it was found that lower nutrient intake of preschool children was correlated with higher dependency ratios [6]. This pattern has also been noted in Jamaica; there, higher dependency ratios were associated with declines in dietary complexity of children’s diets [8]. In addition, a combination of variables indicative of “dependency stress” [8, p. 2501 were found to be correlated with malnutrition in children. Another focus of recent research in intrahousehold dynamics is that of intrahousehold income allocation. Some attention has been focused on the ways working women allocate their incomes, and thus through their earnings improve the diets and nutritional status of their families [9-l 11. What has been neglected is the importance of household role, as opposed to gender; do all working women’s incomes have the same effect, or are incomes of mothers and teen-aged daughters, for example, spent in different ways. Another topic which has not been adequately explored is how many other wage earners are part of the household, and how they spend their earnings [12]. For example, in north-eastern Ghana, everyone over the age of 10, both male and female, has some form of cash income, over which he/she largely has control. Household size tends to be large; 67% of the households consisted of between 5 and 20 adults [13]. Studies in Mexico City also reflect that there is no general pattern for the extent to which grown children help pay for household expenses [14]. The mother’s 305



ability to convince her children to contribute and the economic circumstances of the particular household seemed to be the key variables. Even in cases where children did allocate their income toward household expenses, they typically contributed at most only a portion of their earnings [14]. An important issue, then, is the extent to which household members make all, a portion, or none of their earnings available to those responsible for purchasing household necessities, including food. It has generally been assumed that all workers make all of their incomes available for household expenses, so that no difference would exist between the total household income and the income available for the purchasing of food. If members of a household contribute only a portion of their earnings for use for household needs, the greater the number of workers in the household, the more likely it is that the total amount of income entering a household will differ greatly from the amount that becomes available to those responsible for household purchasing. A large difference would be predicted between total and available incomes. Thus, income allocation patterns, rather than mere household size would be the reason behind the correlations between nutritional problems and larger household size. This paper, then, suggests an alternative approach to understanding intrahousehold allocation of food, which begins with the consideration of patterns of intrahousehold allocation of income. The focus is on a variable we will call ‘available income’. Available income refers to the amount of money which is actually available to those in the household responsible for household expenditures, including food. The concept of available income is not related to who makes decisions about how the income is spent; it merely refers to the sum of money about which decisions on household expenditures are made by those in the household responsible for such decisions. Available income is not the same as ‘total food expenditure’. Instead, it represents the total sum which is available to those who made decisions about all household expenditures, of which expenditures for food are but one part. Food consumption and nutritional status are compared in smaller and larger households in northwestern Mexico. Also considered are other important differences between the two types of households, including total and available incomes, household size, number of economically active household members, and dependency ratios (i.e. the ratio of income earners to dependents). METHODOLOGY Study area

The data on which this paper is based were collected in 1982 in Sonora, a state in northwestern Mexico [15]. This is an extremely arid region, the eastern section (la Sierra) of which is quite moun-

tainous. The study focused on populations from Hermosillo, the state capital, and Arroyo Lindo (a pseudonym), a village in the Sierra region, but at a similar altitude, so that the climates are nearly identical. The two locations were chosen because nutritional surveys had indicated that mild to moderate malnutrition was much more common in those areas than would have been expected in this relatively well developed area of Mexico [ 16, 171. In 1980, Hermosillo had a population of about 281,697 [18]. This city is not heavily industrial, although major employers include food processing companies, and a cement plant. Primarily, Hermosillo serves as a service and shopping center for the villages of the Sierra region to the east, and for the large scale irrigation areas to the west (la Costa). The city has grown tremendously in recent years, largely due to rural to urban migration within Sonora [19], as well as some migration from areas of Mexico further to the south. Arroyo Lindo is a largely cattle ranching community, with both cheese and beef being produced, and sold. Households then use the cash to buy food for their own consumption. The village is located about an hour’s travel from Hermosillo, a factor which has contributed to large scale migration to that city. This pattern has produced a population size which has changed very little in recent years; in 1980, the population, all of which is mestizo, was about 1713 [18]. Sampling

The sample of 78 households was drawn from the populations of Hermosillo and Arroyo Lindo, because studies [16, 171 indicated that nutritional problems existed in those areas. Much of the migration to Hermosillo comes from villages in the Sierra region at higher altitudes than Hermosillo. Resulting differences in temperature and climate might affect food consumption patterns. To hold the effects of climate as constant as possible, it was desired to have the Hermosillo sample comprised either of households of long-term Hermosillo residents or of migrants from Arroyo Lindo. Sampling was, thus, non-random. Twenty households were selected from Arroyo Lindo. A criteria for selection was that one of the household heads (this refers to the husband and wife in each household) had a sibling married to another person from Arroyo Lindo, both of whom had migrated to Hermosillo. The most practical way of finding Arroyo Lindo households of the desired type was to seek referral from informants of different families who lived in different areas of the village. This sampling strategy in the rural area made it possible to identify households to sample in Hermosillo; the rural households provided the addresses of their relatives in Hermosillo. Selection of the rest of the Hermosillo sample was more problematic, because most urban residents were migrants from other parts of the Sierra region. This made households with both the sampling


allocation of resources

criteria somewhat rare. As such, a system of using networks (very similar to the strategy used in Arroyo Lindo) was employed. People who had been born in Hermosillo were asked to suggest families that met the sample criteria. A number of different networks in different areas of the city and from different income strata were employed. Each household sampled had both male and female heads (i.e. husband and wife), and at least two children; this criteria was included so that all households were of a similar type. No household where any woman was more than four months pregnant or lactating was included, to avoid individuals whose nutritional needs would be hard to quantify. For this same reason, only individuals between 1 and 60 years were included in the calculation of the means presented below. While the total sample includes households of a wide range of incomes, for some of the analyses presented, the groups were divided into lower (< 49,000 pesos) and higher ( > 49,000 pesos) income groupings of per capita annual income. This point represents a natural break in the distribution of incomes, as well as being roughly analogous to a per capita income in the United States of $4000 per year. This is midway between the amount estimated to represent food stamp eligibility (about $3000 per capita per year) and middle income status (about $5000/per capita per year) in Tucson, Arizona, the closest U.S. city to Sonora. The value of the peso at the time of the study was equal to U.S. $0.025. Data colIection and anaIysis

Standard anthropological research techniques of participant observation were used. The first author lived with families in the communities being studied, and was able to observe daily diets and food habits. An interview on aspects of household economics and food utilization was conducted with the female head of each household, because she is the person who generally had responsibility in these areas. In addition, food consumption data were collected by


quantitative recall for each individual in the household for two weekdays, as well as for Sunday, a day on which the diet tends to differ from the rest of the week. Each item eaten was recorded as whether it was consumed at home or away from home. Each individual was asked to recall all food consumption for the previous 24 hr, and to indicate, using measuring cups, spoons, and other measures, how much of each item was eaten. Mothers were asked about food which they served to their children, and queried carefully about how much of the food that was served to the child was actually consumed. The children themselves were asked, when at all possible, about snacks and other food consumption. If a child was too young or otherwise unable to do this, the older child or adult who was with the child at the time the food was consumed was asked to provide this information. Mean daily consumption of 10 food types was calculated as standard servings, adapted from the Michigan Department of Public Health [20] exchange system (see below). The 10 food types chosen were based on the basic food groups, expanded to cover the range of culturally important food categories in this area. The food types and standard serving units are given in Table 1. The standard servings were converted into a daily average for each individual by multiplying the standard servings consumed on each of the weekdays by 3, adding Sunday’s consumption, and dividing the total by 7. All of the children 13 years and under in the Arroyo Lindo households sampled were weighed and measured. It was not possible logistically to collect these data on the children from the urban households due to the inaccessability of some of the homes to a road, and the weight of the scale. The first author weighed and measured the children at the village health center, with the assistance of the local doctor and/or nurse. The scale was later checked with an Accuwiegh Scale and found to be accurate. The total

Table 1. Food types and standard serving units Food tvue

Descrietion Oranges, watermelon, bananas, limes, quince

Vegetables Beans Dairy

Eggs Grains Miscellaneous

Potatoes Meat Alcohol

Zucchini, lettuce, cabbage, onions, carrots, tomatoes, green Chile, red Chile, peas, Chile sauce, spinach and other greens, fresh coriander, cucumbers Pinto beans, garbanzo beans, lentils Fresh milk, evaporated milk, powdered milk, regional cheese (similar to cottage cheese), canned cream, Chihuahua cheese (similar to Monterey jack), asadero cheese (similar to mozzarella), ice cream, yogurt Chicken eggs Flour tortillas, corn tortillas, bread, all varieties of pasta products, pan dolce (sweet rolls) rice, cake, crackers, cold and cooked cereal Soda, candy, Kool-Aid, potato chips (and related foods, i.e. fritos, doritos, etc.) White potatoes Beef and pork (both with very little fat trimmed), chicken, tuna, coldcuts, sausage, fish, turtle, very little of variety meats other than menudo (tripe) and cabeza (beef head) Beer, mixed drinks

1 unit eauals f c juice or cooked/canned or 1 medium piece i c juice or cooked/canned or j c raw f c cooked 1 c milk or 1 c plain yogurt or If oz aged cheese I medium egg 1 slice bread or i c cooked cereal, spagetti, noodles 1 soft drink (473 ml) or 1 small package potato chips, etc. or I medium sized candy bar or 1 c Kool-Aide 1 medium potato 2-3 oz cooked meat/fish

I can beer or 1 mixed drink or




number of children weighed was 75. The Ramos Galvan Standards [21], which were developed for Mexico were used to evaluate the anthropometric data. The ages used were those given by the mothers, and were to the month for children under 2 years, and to the year for all other children. Income data were collected through ethnographic interviewing on various aspects of the topic of income allocation, including, how much each worker contributed to either the husband or wife, how the income of each household member was used, how decisions were made on household expenditures, and how increased income would be spent. Total income refers to the total amount of income earned by all workers in the household. Available income is a calculated variable, and refers to the total amount of money that was received by the husband and/or wife, i.e. the total of each of their earnings, plus whatever contributions they received from other working members of their households. All income data given, whether referring to total or available incomes, represent mean per capita annual incomes. RESULTS

Household size: general

The data from this study largely corroborate the negative relationships reported in the literature between larger household size and food consumption. Households smaller than or equal to the regional mean-6 persons per household [16, 1‘II-consume greater amounts of all food types per capita for which there are significant differences, except beans and grains (Table 2). This is more likely due to the effects of income than household size. Table 3 shows a comparison of the total and available per capita incomes by household size. Total and available incomes are not significantly different for the smaller Table 2. A comparison of mea” daily per capita servings in small and larae households

Food type Fruit Vegetables Beans

Small HH mea” N = I65 indiv. (36 hh) 1.2 range: &8 0.9 range: &5 1.0 range: &5


1.4 range: &8

Eggs Grains Miscellaneous Potatoes Meat Alcohol

I.0 range: I%3 5.4 range &I 2 1.2 range: O-5 0.4 range: O-3 1.7 range: o-8 0.1 range: O-4

Large HH mea” N = 354 indiv. (42 hh)
















range: i&6

0.8 range: &4 1.8 range: 0- I3 1.0 range: &7 0.9 range: 04 6.2 range: &23 1.3 range: (t5 0.4 range: o-3 1.2 range: G-5 0.2 range: C-l I



I .34






Table 3. A comparison of total and available per capita income per household size Household size


Small households ( Q 6) Mean SD Range L! 66 0.48


P Larger household (> 6) Mean SD Range k 611.76 P

Total income

Available income

52,474 36,238 12,920-227,235

49,255 36,859 7154227.235

42 48,828 29,003 9984-l 12,857

36,317 25,987 2362-104.000


households, whereas they do differ significantly for the larger households (using a one-tailed test, since available income cannot be expected to be greater than total income). Dependency ratios are mentioned in the literature [6-81 as being a factor which may also be of importance in explaining the food consumption problems observed in larger households. However, in this case, no significant differences were observed in dependency ratios between the larger and smaller households (3.62 vs 3.35, t = 0.64, df = 65, ns). The larger households do not have more dependent children; they have more adults (and young adults), and more adults to be economically active. The smaller households, then, are not only smaller, but are structurally different from the larger households. The smaller households tend to be of a type in which the only household incomes are those earned by the husband and/or wife, as opposed to the multiple worker pattern which is more common in the larger households (Table 4). This structural difference will have the effect of making per capita total and available incomes in the smaller households more similar than is the case in the larger households. The different structure of the larger households (and its critical implication for income allocation) has the effect of making per capita total and available incomes more different than is the case in the smaller households. In households of lower per capita available incomes (< 49,000 pesospthose which would be predicted to be at greater nutritional risk-per capita food Table 4. The association between household size and number of economicallv-active members Only heads of household economically active

Multiple workers

Household size





Small households (<6) Large households (>6) Total











x2 = 16.399 [email protected]



Intrahousehold allocation of resources consumption of the smaller households differs significantly from that of the larger households for vegetables, miscellaneous, meat, and alcohol, with the smaller households consuming greater per capita amounts of these foods. Therefore, the problems of the larger household appear to be related to its lower per capita available income and thus, lower per capita consumption of almost all food types. An important question must be addressed at this point with regard to expectations of contributions to household expenses by workers other than the household heads. In a situation of post-marital residence of young couples with parents (usually following a patrilineal pattern), which is not an uncommon pattern in this area, lack of contributions to household expenses is a way of enabling these couples to save money in order to set up their own households. For primarily adolescent children and/or unmarried young adults, expectations as to amount of contributions to household expenses might be much greater, although the earnings of these workers might be theoretically predicted to be quite low, due to their ages. In the households in this study, the “young couple living with parents but saving for their own household” pattern was rare. Of a total of 55 non-household head workers, 8 were the husband and/or wife of a ‘young couple’. If the single mothers supporting children, but living with their parents are also included, the total of those who might have good reason for controlling all of their own incomes is 11 individuals. 80% of the non-household head workers in these households were adolescent children, and/or unmarried adults, of whom contributions of a more substantial nature might be expected. Further, due to increased education among the younger generation, the actual wages of many of these unmarried workers were quite good. Regardless of gender of the worker, 11 (25%) of them earned more than the male head of the household in which they resided, 8 (18%) earned the same, and 25 (57%) earned less. Household size: relationship to nutritional status in Arroyo Lindo

The connection between household size, number of economically active members of a household, total and available incomes, and food consumption with nutritional status can be seen by examining these patterns in the 20 Arroyo Lindo households. If a household had more than one child who was at least 1st degree malnourished in terms of height for age (i.e. ~95% of the standard [21]), the household was classified ‘stunted’. Households with more than one child who was at least 1st degree malnourished in terms of weight for height (i.e. ~90% of the standard [21]), were classified ‘wasted’. The remainder of the households were considered as ‘normal’. Because of the relatively small number of households and children involved, these data were not analyzed for different patterns due to gender.


Table 5. Number of economically active members of ‘stunted’ and ‘normal’ Arrovo Lindo households Only beads of household economically active Tyue of household Stunted Normal

N = 10 N=lO

Multiple workers





4 9

40 90

6 1

60 10

x2 (Yate’s Correction) = 5.71. df= I. P C 0.025.

For the Arroyo Lindo households, as was the case for the total sample, per capita available incomes were lower for larger households, as was per capita dietary intake. The household structural pattern in which there are multiple workers in addition to the household heads was also more common in the larger Arroyo Lindo households. When the stunted households were compared with the normal ones, household size was significantly different in the stunted households-10.3 vs 6.0 for the normal households (t = 3.5, df= 18, P < 0.01). The magnitude of the differences in households sizes far surpasses that which would be due to the bias introduced by the methods of classification of the households (i.e. the way in which non-normal households were defined makes it somewhat more likely that a larger household-ones with more children-will have the chance to ‘express’ malnourished children, and therefore not be classified as ‘normal’ than will a smaller household). Stunted households were also more likely to be of the type in which there are multiple workers in addition to the household heads (Table 5). Here, as in the sample as a whole, these findings are related to available incomes (Table 6), which drop greatly below those of total household income only for the ‘stunted’ households. While these differences are not statistically significant for this small subsample, the pattern does seem to follow that of the larger sample (Table 3) for which significant relationships were observed. Patterns of food consumption for this subsample also Seem to follow that observed in the larger sample (Table 2). Food consumption was lower for the ‘stunted’ households as compared with the ‘normal’ households for all food types for which there were Table 6. A comparison of total and available per capita income in ‘stunted’ and ‘normal’ Arroyo Lindo households Household classification Stunted Mean SD

0.56 d; 18 P ns N=lO Normal Mean SD


d;. P

Total income

Available income

34,589 2 I ,495

24,738 15,748

39,962 21,057

38,818 21,458

N= 10

18 “P




Table 7. A comparison of mean daily per capita servings in Arroyo Lindo ‘stunted’ and ‘normal’ households



Food type

Stunted HH mean N =9X indiv. (IOhh)

Normal HH mean N = 52 indiv. (IOhh)



Fruit Vegetables Beans Dairy JW Grains

0.4 0.7 1.7 0.7 IO 6.5

Miscellaneous Potatoes Meat

I.? 0.4 0.9



I.1 0.8 2.1 0.9 I.? 6.6 1.4 0.6 1.0 0.2

-3.14 -0.39 - 1.43 -1.11 ~ 2.05 -0.22 - 1.42 -3.53 --0.49 -0.01

significant differences-fruit, dairy, and potatoes (Table 7). Similar patterns were seen when the ‘wasted’ households were compared with the ‘normal’ households. CONCLUSIONS

These data suggest that the nutritional problems of large households may be related to the low available incomes characteristic of these situations. Much of the correlation cited in the literature between household size and nutritional problems may actually be due to differences in per capita income among the households. When income is controlled, the Sonoran data suggest that while per capita dietary intakes of most food types are lower in larger households, so also are per capita available incomes. This pattern is related to the frequency with which larger households are of the multiple worker type, in which available incomes tend to be much lower than in households in which only the husband and/or wife are working. Consideration of these issues for the Arroyo Lindo subsample for which anthropometric data were available suggest that the result of the lower available incomes seen in the larger households often may be malnourished children. These data indicate the importance of household income allocation patterns in the determination of household food consumption patterns. The larger household may appear to be better off in terms of total household income, but nutritional status is likely to be less positive than anticipated. Low dependency ratios do not compensate for the lower amounts of income available to those purchasing household necessities, including food. Differences in available income are thus, a critical factor to consider in predicting the nutritional well-being of a given household.

















18. 19.

20. Acknowledgements-This research was funded by a grant from the International Food Policy Research Institute. Additional support was provided by the Comins Fund of the University of Arizona, and the Centro de Investigation in Alimentacion y Desarrollo (Hermosillo, Sonora)


Pinstrup-Andersen P. Estimating the nutritional impact of food policies: a note on the analytical approach. Paper presented at the XI International Congress of the International Union of Anthropological Sciences, Vancouver, 1983. Pelto G. Intrahousehold food distribution patterns. In Malnutrition: Determinants and Consequences, pp. 285-293. Liss, New York, 1984. Wolfe B. C. and Behrman J. K. Determinants of child mortality, health, and nutrition in a developing country. J. Dec. Econ. 11, 163-193, 1982. Whiteford M. The interface of cultural factors with nutritional well being. Med. Anthropoi. 6, 221-230, 1982. Wray J. D. and Aguirre A. Protein-calorie malnutrition in Candelaria, Colombia. J. trop. fediat. 15, 76698. 1969. Chaudhury R. Determinants of dietary intake and dietary adequacy for pre-school children in Bangladesh. Food Nutrit. BUN. 6, 24-33, 1984. Chaudhury R. Adequacy of child dietary intake relative to that of other family members. Food Nutrit. Bull. 10, 26.-34, 1988. Marchione T. Factors associated with malnutrition in the children of Western Jamica. In Nutritional Anthropology (Edited by Jerome N. et al.), pp. 2233213. Redgrave, Pleasantville, 1988. Guy& J. Household Budgets and Women’s Incomes, Working Paner No. 28. African Studies Center. Boston Univers%y, Brookline, 1980. Dwyer D. Women and Income in the Third World: Implications for Policy, Working Paper No. 18. The Population Council, New York, 1983. Tripp R. Farmers and traders: some economic determinants of nutritional status in Northern Ghana. Food Nutrit. Bull. 8, 3-11, 1982. Baer R. D. Inter and intrahousehold income allocation: implications for third world food policy. In Human Dimensions of Food Policy (Edited by Macmillian D.). University of Georgia Press, Athens, 1990. Whitehead A. “I’m hungry, Mum”: the politics of domestic budgeting. In Of Marriage and the Market (Edited by Young K. et al.). CSE Books, London, 1981. Beneria L. and Roldan M. The Crossroads qf Class and Gender. The University of Chicago Press, Chicago, 1987. Baer R. D. The interaction of social and cultural factors affecting dietary patterns in rural and urban Sonora. Mexico. Doctoral Dissertation, University of Arizona. University Microfilms, Ann Arbor, 1984. Valencia M. Estudio nutritional en la zona Serrana de1 Estado de Sonora. Instituto de Investigaciones y Estudios Superiores del Noroeste. A. C. Hermosillo, Sonora, 1980. Typescript. Valencia M. Nutritional en centros urbanos marginados de Sonora. Instituto de Investigaciones y Estudios superiores del Noroeste. A. C. Hermosillo, Sonora, 1981. Typescript. Estado de Sonora X Censo General de Poblacion J Mcienda: Resultados Prelimittares. Mexico, D.F., 1980. Baer R. D. Land tenure and agricultural strategy__ as constraints on the household production of human nutrition. Food Nutrit. Bull. 12, 279-288. 1990. Michigan Department of Public Health Basic Nutrition Facts. Michigan Department of Public Health, Lansing, 1980. Ramos Galvan R. Archives de Investigation Medica. Somatometria Pediatrica No. 6, Suppl. 1, 312-390, Ediciones Culturales Mexicana, S.A., Mexico, 1975.