Skip Navigation


Age and Ageing Advance Access originally published online on March 19, 2007
Age and Ageing 2007 36(3):286-291; doi:10.1093/ageing/afm016
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
36/3/286    most recent
afm016v1
Right arrow E-letters: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when E-letters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Holsinger, T.
Right arrow Articles by Plassman, B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Holsinger, T.
Right arrow Articles by Plassman, B.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Copyright © The Author 2007. Published by Oxford University Press on behalf of the British Geriatrics Society.

Intelligence in early adulthood and life span up to 65 years later in male elderly twins

Tracey Holsinger1,2,, Michael Helms2 and Brenda Plassman2

1 Durham VA Medical Center, Psychiatry, Durham, NC, USA
2 Duke University Medical Center, Psychiatry, Durham, NC, USA

Address correspondence to: Tracey Holsinger. Tel: (919)286-6933; Fax: (919)416-5832. Email: tracey.holsinger{at}med.va.gov


   Abstract
Top
 Abstract
Introduction
 Materials and methods
 Results
 Discussion
 Key points
 References
 
Background: previous research has reported that greater intelligence in early life is associated with longer lifespan. Whether this relationship is mediated by genetic factors or environmental factors, some of which could be modified by an individual, is unclear.

Objective: we examined the association between intelligence test scores, obtained during the 1940s, and age at death in a group of 492 male twin pairs, members of the National Academy of Sciences - National Research Council Twins Registry of WWII veterans.

Design: using self-report information collected in th 1960s, we xamined whether modifiable risk factors for mortality, such as use of tobacco and alcohol, cardiovascular disease, and body mass index altered the association between intelligence and longevity.

Results: when each member of a twin pair was treated as an independent observation, higher intelligence test scores were associated with longer life span (P = 0.0002). Modifiable risk factors were associated with life span as expected. However, in co-twin control analyses in which one twin served as the control for the other twin, neither intelligence nor any modifiable risk factors showed a significant association with life span.

Conclusion: our findings suggest that genetics and early life environmental factors contribute heavily to lifespan and when one controls for these factors using twins, the effect of intelligence on longevity is diminished.

Keywords: intelligence, survival, twins, risk factors, elderly


   Introduction
Top
 Abstract
 Introduction
Materials and methods
 Results
 Discussion
 Key points
 References
 
Previous studies have reported that higher scores on intelligence tests in childhood and young adulthood may predict a longer life span [1–4]. These findings prompted a few researchers to investigate whether the effect of intelligence or educational level on longevity is altered by other well-known risk factors of premature mortality such as tobacco use and cardiovascular disease [5, 6]. The general conclusion from these latter studies is that the effect of intellect is fairly independent of the effect of health behaviours and medical conditions on mortality. However, these studies have not assessed the extent to which genetic influences may explain some of the association between these risk variables and mortality.

Longevity has a genetic component. Estimates of the heritability of life span, as measured in twins, have varied somewhat but are generally estimated around 0.25 [7–9]. This might be due to the effects of specific genes or might be mediated by genetic influences on life-shortening conditions or life-style factors, such as socio-economic status, cardiovascular disease, smoking and body mass index (BMI). Some of these factors, such as smoking, BMI and cardiovascular disease, themselves have been shown to have genetic components [10, 11].

It is possible that the same latent genetic or environmental influences underlie both these various medical conditions and longevity, and that a lack of understanding of such confounding may result in erroneous inferences about the causality in the relationship between such factors and longevity. In the case of BMI and smoking, there does not appear to be an overlap between the genetic and shared environmental influence on these factors and longevity [12]. However, little is known about the overlap in aetiology of some other conditions, such as cardiovascular disease. Given that intelligence also has a genetic component [13, 14], one might ask whether the same issue of potential confounding may explain the reported association between intelligence and longevity.

We examined this question using twins. The study of twins inherently controls for many genetic and early life shared environmental influences. Using the National Academy of Sciences - National Research Council (NAS-NRC) Twin Registry of WWII Veterans from the United States, we examined the association between intelligence in early adulthood and longevity approximately through age 80. We also investigated whether smoking, alcohol use, BMI, and other cardiovascular risk in midlife contributed to mortality, and whether any of these factors altered the association between intelligence and longevity. Finding an association between early life intellect and longevity (independent of the other factors) within-twin pairs who share not only many (or all) of the same genes, but also childhood socio-economic status, would provide support to the unique contribution of IQ to life span.


   Materials and methods
Top
 Abstract
 Introduction
 Materials and methods
Results
 Discussion
 Key points
 References
 
Subjects
The subjects were members of the NAS-NRC Twin Registry of WWII Veterans born between 1917 and 1927. To be included in the study, both members of the twin pair had to have a military entrance examination score available and had to have completed a questionnaire that was mailed to Twin Registry members in 1967 to collect information on a number of medical conditions and health behaviours. The resulting cohort included 492 twin pairs who had lived at least into their 40s, of which 301 were monozygotic and 191 were dizygotic. Of these, as of 31 December 2004, both members were deceased in 115 pairs, one twin was deceased in 155 pairs, and both twins were still alive in the remaining 222 pairs. In 72 pairs, zygosity was determined from buccal and blood DNA samples. In 420 pairs, zygosity was determined from the NAS-NRC Twins Registry data based on the best available information from questionnaire responses, fingerprint analysis, and anthropometric data from military records [15, 16]. Methods of establishing zygosity are estimated by cross-validation to be 97% correct [17].

To obtain as complete data as possible, death information was determined from all available sources: (i) the computer-based Beneficiary Identification and Records Locator Subsystem of the Veterans Administration, (ii) telephone calls with the subject's family, or (iii) the Social Security Death Index database.

Measure of intelligence
As part of the entrance procedures for military service during the 1940s, all army inductees and enrollees in officers candidate school were administered the Army General Classification Test (AGCT) [18], while those entering the navy were administered the General Classification Test (GCT) [19]. The AGCT was developed to be a ‘test of general learning ability’ and the GCT were designed to measure verbal aptitude. Both tests correlate with education (r = 0.65–0.73) [20, 21] and with other tests that assess or are correlated with general intellect. For example, results of the AGCT have correlated strongly (r = 0.83) with results of the Wechsler–Bellevue Intelligence Scale, which measures verbal and nonverbal factors of intelligence [22].

Test scores for the AGCT and the GCT were standardised to a common metric with a mean of 10 and standard deviation of 1. Assumptions underlying this common standardisation were that the two variable's population distributions were, in fact, very similar in shape and would not have differed in their mean scale measures had both been administered a common scale. We jointly refer to the scores from these tests as GCT/AGCT scores.

Mailed questionnaires
Information on a number of medical conditions and health behaviours was collected on a questionnaire mailed to the registry members in 1967. For this study, we selected life-style factors or medical conditions that previously have been reported to be associated with longevity, but also have the potential to be modified by the individual and are likely influenced by genes. The factors were defined as follows. BMI was calculated from self-reported height and weight. Smoking history required that the individual be a current smoker (at the time of the 1967 survey) and then was defined by the number of pack-years smoked. Heavy alcohol use was defined as any of the following: reported daily consumption of more than three beers, more than one-half bottle of wine or more than three drinks of liquor or reported daily hangover or drinking to intoxication daily.

The Rose Questionnaire, developed as a tool to diagnose angina pectoris in epidemiological studies [23], was included in the mailed questionnaire. A positive Rose Questionnaire required a report: (i) of pain in either the sternum or left arm and left anterior chest, (ii) provoked by exertion, (iii) requiring a lessening of exertion or use of nitroglycerin, and (iv) pain disappearing within 10 min when subject is standing still. We categorised the responses on the Rose Questionnaire in two ways; first as a dichotomous variable (i.e. positive versus negative) as it was designed and then as a continuous variable from 0 to 4, representing the sum of the Rose items endorsed to capture all available information. We created an additional variable ranging from 0 to 3, referred to as a vascular score, to reflect the severity of cardiovascular disease. This variable included one point each for a positive Rose Questionnaire, a history of severe chest pain, and a history of myocardial infarction.

Data analysis
Means, standard deviations, and frequencies were calculated. Heritability, the ratio of genetic variation to total variation, was calculated for GCT/AGCT scores, age at death, a positive Rose Questionnaire, the variable indicating number of positive responses on the Rose Questionnaire, heavy drinking and smoking variables using Falconer's estimate, h2 = 2 (rMZ—rDZ), that is, twice the difference in the pair correlations in MZ and DZ pairs. Pearson correlations were used to characterise continuous variables and polychoric correlations were used for dichotomous or ordinal variables with limited levels. Cox proportional hazards models were used to model and test the relationship between GCT/AGCT scores and the time from completing the mailed questionnaire to the date of death or the censoring date, 31 December 2004, while adjusting for age at completion of questionnaire. All models included a dummy variable denoting the source of the test score that was coded 1 for AGCT and 0 for GCT. Covariates of BMI, smoking habits, alcohol use and several cardiac risk variables were also modelled singly and in concert with GCT/AGCT scores. Additionally, models containing the interaction of GCT/AGCT and each of the other variables were investigated to determine if they modified the association between GCT/AGCT and lifespan as either independent modifiers of mortality or as confounders through a function of the GCT/AGCT.

Two classes of models were considered and fitted. Both used the same set of twin pairs, but in one set the individuals were analysed as independent observations (i.e. case-control analyses), while in the second they were considered in a manner dependent on the pairing, through stratification by twin pairs (i.e. co-twin control analyses). In the independent models (i.e. case-control analyses), we adjusted the covariance matrix of the beta coefficients to partially account for the dependencies between members of a twin pair, utilising a previously published SAS macro [24].


   Results
Top
 Abstract
 Introduction
 Materials and methods
 Results
Discussion
 Key points
 References
 
Table 1 provides descriptive statistics grouped by living and deceased status. Information regarding whether the twins had lived together during childhood was collected as part of the mailed questionnaire and indicated that all but one pair had lived their first 15 years together. Table 2 shows heritability estimates for GCT/AGCT test scores and many of the health variables. Although GCT/AGCT showed substantial heritability, scores on this measure for two members of a twin pair differed on average by 0.55 standardised units (SD = 0.58). For twin pairs in which one twin had died, the mean difference between age at death for that twin and age of censoring for the living twin was 6.7 (SD = 3.7) years. In the twin pairs in which both twins had died, the mean difference between ages of death was 7.7 (SD = 6.5) years.


View this table:
[in this window]
[in a new window]

  		Table 1. Characteristics of twins

 


View this table:
[in this window]
[in a new window]

  		Table 2. Intra-pair correlation coefficients by zygosity

 
Case-control analyses
Figure 1 shows that when subjects were analysed as individuals (not members of twin pairs), a longer life span was associated with higher intelligence (hazard ratio (HR) = 0.698 per 2 standard units, P = 0.0013). As expected, smoking history, heavy alcohol use, symptoms suggestive of vascular disease both from the Rose Questionnaire and the vascular score, were all correlated with life span. In models that included GCT/AGCT score, source of test score, and age at questionnaire as covariates, the following relationships were found. Smoking, as measured in pack-years, was correlated with increased risk of death (HR 1.29, P < 0.0001). Heavy alcohol use was associated with shortened life span (HR 2.26, P < 0.0001). Coronary artery disease as measured by a positive Rose Questionnaire was associated with an increased risk of dying (HR 1.67, P = 0.045). Using the items of the Rose Questionnaire summed was also associated with an increased risk of dying (HR 1.52, P = 0.009). A higher vascular score was also associated with an increased risk of dying (HR 1.76, P < 0.0001). Based on previous literature [25], we examined the linear and quadratic function for body mass index and found no association with longevity.


Figure 1
View larger version (27K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 1. Kaplan Meier survival curves for four categorised levels of the AGCT/GCT test scores.

 
No significant interactions were found for smoking, heavy alcohol use or any vascular scores with IQ as measured in early adulthood. Including these variables in models with the variable for intelligence did not change the effect of IQ on life span. The dummy variable denoting the source of the test score (i.e. AGCT or GCT) was never significant in the models (most significant probability = 0.19, but generally it was >0.8).

Co-twin control analyses
In the co-twin control model conditional on twin pairing, IQ was no longer associated with longevity (HR = 1.06, P = 0.84). To explore whether the intra-pair similarity on GCT/AGCT was limiting our ability to detect an association, we then re-ran the analyses limiting the sample to only twin pairs for whom the within-pair difference on the IQ testing was greater than the mean intra-pair difference for the entire group (360 pairs). In this analysis, the effect of test scores on life span was stronger but still did not approach statistical significance (P = 0.59).

In co-twin control models using the entire sample, smoking history (HR = 1.11, P = 0.10), heavy alcohol use (HR = 1.39, P = 0.23), BMI (HR = 0.88, P = 0.34), and the vascular variables (HR = 1.13, P = 0.59 for a positive Rose Questionnaire, HR = 1.21, P = 0.67) for summed Rose Questionnaire variable, HR = 0.96, P = 0.94 for our composite vascular variable) were not associated with age at death. All co-twin control models were also run separately by zygosity and none of the results reached statistical significance either for monozygotic or dizygotic twins.

In the co-twin control analyses, only the twin pairs in which one or both members of the pair were deceased were informative. In twin pairs where both members survive, no events (death) occur within-twin strata and they do not contribute information to the analysis. However, the reduced number of informative individuals did not seem to explain the lack of significant findings, as the proportional hazard regression coefficient also decreased between the independent analyses and the co-twin control analyses.


   Discussion
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
Key points
 References
 
In case-control analyses, higher IQ predicted longer life span through approximately age 80 years. It is striking that the magnitude of the association between IQ and life span reported in the present study is close to that reported by a previous study, that is, a relative risk of 0.68 for a two-standard deviation change for men [3]. The presence of modifiable risk factors in midlife such as smoking, heavy alcohol use and vascular disease also impacted life span in predictable ways, but did not alter the association between IQ and longevity. In contrast, in co-twin control analyses where each twin was used as a matched control for his co-twin, neither the differences in IQ nor a modifiable risk factors had an appreciable effect on life span.

To our knowledge, this is the first study to investigate in twins the reported association between early life intellect and life span. The discrepant outcomes from our case-control and co-twin control analyses suggest that IQ alone may have little impact on longevity. Rather, higher scores on intelligence tests may reflect the combination of genetic predisposition and healthy early environment—contributing to a higher level of ‘fitness’ of both the brain and body that results in a longer life span [3]. We note, however, that these factors are largely shared in twin pairs, leading to reduced variation within pairs compared to biologically unrelated individuals. Although the within-twin pair differences in AGT/AGCT scores in our sample averaged about one-half of a standard deviation; we cannot exclude the possibility that this somewhat limited variability was insufficient to detect an impact on life span.

A few other twin studies have investigated whether the genetic influences on mortality are accounted for by the genetic influences on covariates such as smoking, BMI, heavy alcohol use and vigorous physical exercise [9, 10, 12]. Their findings suggest that there is little overlap between the genetic influences on smoking and BMI and mortality, but the findings for heavy alcohol use, physical activity and mortality are less clear-cut. It is noteworthy that these studies detected an association between these modifiable risk factors and mortality in co-twin control analyses, suggesting that limited within-twin pair variance does not explain our inability to find such an association in the present study. These previous studies suggest that multiple genetic and non-genetic factors contribute to life span. It is possible that analyses without the sort of genetic control provided by twins could result in over-interpretation of associations between IQ and life span. Our findings suggest that genetics and early life environmental factors contribute heavily to life span and when one controls these factors by using twins, the effect of intelligence is diminished. However, we acknowledge that, due to the within-twin pair similarity for many of these variables, the criteria for detecting differences is more stringent in the co-twin control design.

Limitations of our study include the right censoring of many of the veterans because they are still alive. It is possible that fewer censored individuals would alter the results somewhat; however, given that the co-twin control analyses typically did not approach significance, it is unlikely that this would alter the general conclusions. Another limitation is the use of self-reported risk factor information, which generally leads to an under-reporting of the risk factors. However, it seems unlikely that this would be limited to the veterans who lived longer and thus, would not explain our results.


   Key points
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Key points
References
 

  • When twin pairs are analysed as individuals, modifiable risk factors affect longevity as expected.
  • When twin pairs are analysed as individuals, higher intelligence in young adulthood is associated positively with longevity.
  • When twin pairs are analysed as pairs, neither modifiable risk factors or intelligence is associated with longevity.
  • Genetic factors and early childhood environment may weigh heavily on longevity.


   Acknowledgements
 
Supported by a grant from NIH AG-08549 (Plassman). The authors gratefully acknowledge the technical and scientific contributions of Dr William Page and Ms Harriet Crawford at the Medical Follow-up Agency.


   References
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Key points
 References
 

  1. O'Toole BI, Stankov L. Ultimate validity of psychological tests. Person Individ Diff (1992) 13:699–716.[CrossRef]
  2. Hart CL, Taylor MD, Davey Smith G, Whalley LJ, Starr JM, Hole DJ. Childhood IQ, social class, deprivation, and their relationships with mortality and morbidity risk in later life: prospective observational study linking the Scottish mental survey 1932 and the midspan studies. Psychosom Med (2003) 65:877–83.[Abstract/Free Full Text]
  3. Whalley LJ, Deary IJ. Longitudinal cohort study of childhood IQ and survival up to age 76. BMJ (2001) 322:819.[Abstract/Free Full Text]
  4. Martin LT, Kubzansky LD. Childhood cognitive performance and risk of mortality: a prospective cohort study of gifted individuals. Am J Epidemiol (2005) 162:887–90.[Abstract/Free Full Text]
  5. Hart CL, Taylor MD, Smith GD, Whalley LJ, Starr JM, Hole DJ. Childhood IQ and all-cause mortality before and after age 65: prospective observational study linking the Scottish mental survey 1932 and the Midspan studies. Br J Health Psychol (2005) 10:153–65.[CrossRef][Web of Science][Medline]
  6. Schnohr C, Hojbjerre L, Riegels M, Ledet L, Larsen T, Schultz-Larsen K. Does educational level influence the effects of smoking, alcohol, physical activity, and obesity on mortality? A prospective population study. Scand J Public Health (2004) 32:250–6.[CrossRef][Web of Science][Medline]
  7. Ljungquist B, Berg S, Lanke J, McClearn GE, Pedersen NL. The effect of genetic factors for longevity: a comparison of identical and fraternal twins in the Swedish twin registry. J Gerontol A Biol Sci Med Sci (1998) 53:M441–6.[Abstract]
  8. Herskind AM, McGue M, Holm NV, Sorensen TI, Harvald B, Vaupel JW. The heritability of human longevity: a population-based study of 2872 Danish twin pairs born 1870–1900. Hum Genet (1996) 97:319–23.[Web of Science][Medline]
  9. Hrubec Z, Neel JV. Familial factors in early deaths: twins followed 30 years to ages 51–61 in 1978. Hum Genet (1981) 59:39–46.[CrossRef][Web of Science][Medline]
  10. Kujala UM, Kaprio J, Koskenvuo M. Modifiable risk factors as predictors of all-cause mortality: the roles of genetics and childhood environment. Am J Epidemiol (2002) 156:985–93.[Abstract/Free Full Text]
  11. Kendler KS, Thornton LM, Pedersen NL. Tobacco consumption in Swedish twins reared apart and reared together. Arch Gen Psychiatry (2000) 57:886–92.[Abstract/Free Full Text]
  12. Herskind AM, McGue M, Iachine IA, Holm N, Sorensen TI, Harvald B. Untangling genetic influences on smoking, body mass index and longevity: a multivariate study of 2464 Danish twins followed for 28 years. Hum Genet (1996) 98:467–75.[CrossRef][Web of Science][Medline]
  13. Bartels M, Rietveld MJ, Van Baal GC, Boomsma DI. Genetic and environmental influences on the development of intelligence. Behav Genet (2002) 32:237–49.[CrossRef][Web of Science][Medline]
  14. Plomin R, Spinath FM. Intelligence: genetics, genes, and genomics. J Pers Soc Psychol (2004) 86:112–29.[CrossRef][Web of Science][Medline]
  15. Jablon S, Neel JV, Gershowitz H, Atkinson GF. The NAS-NRC twin panel: methods of construction of the panel, zygosity diagnosis and proposed use. Am J Hum Genet (1967) 19:133–61.[Web of Science][Medline]
  16. Hrubec Z, Neel JV. Twin Research. In: Proceedings of the Second International Congresson Twin Studies—Nance WE, ed. (1978) August 29-September 1, 1977: Washington, D.C. New York: Alan R. Liss. 153–72. The national academy of sciences-national research council twin registry: ten years of operation. Part B. Biology and Epidemiology, Progress in Clinical and Biological Research Volume 24B.
  17. Reed T, Plassman BL, Tanner CM, Dick DM, Rinehart SA, Nichols WC. Verification of self-report of zygosity determined via DNA testing in a subset of the NAS-NRC twin registry 40 years later. Twin Res Hum Genet (2005) 8:362–67.[CrossRef][Web of Science][Medline]
  18. Office UAAGs. The army general classification test, with special reference to the construction and standardization of forms 1a and 1b. J Educ Psychol (1947) 38:385–420.[CrossRef][Medline]
  19. Section SotTaR. Psychological test construction and research in the bureau of naval personnel: development of the basic test battery for enlisted personnel. Psychol Bull (1945) 42:561–71.[CrossRef][Web of Science]
  20. Plassman BL, Welsh KA, Helms M, Brandt J, Page WF, Breitner JC. Intelligence and education as predictors of cognitive state in late life: a 50-year follow-up. Neurology (1995) 45:1446–50.[Abstract/Free Full Text]
  21. Section UASPR. The army generally classification test. Psychol Bull (1945) 42:760–8.[CrossRef][Web of Science]
  22. Tamminen A. A comparison of the army general classification test and the Wechsler-Bellevue intelligence scales. Educ Psychol Meas (1951) 42:646–55.
  23. Rose GA. The diagnosis of ischaemic heart pain and intermittent claudication in field surveys. Bull World Health Organ (1962) 27:645–58.[Web of Science][Medline]
  24. Allison P. Survival Analysis Using the SAS System: A Practical Guide (1995) Cary, NC: SAS Publishing.
  25. Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA (2005) 293:1861–67.[Abstract/Free Full Text]
Received 10 May 2006; accepted in revised form 29 December 2006.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 PediatricsHome page
M. Jokela, G. D. Batty, I. J. Deary, C. R. Gale, and M. Kivimaki
Low Childhood IQ and Early Adult Mortality: The Role of Explanatory Factors in the 1958 British Birth Cohort
Pediatrics, September 1, 2009; 124(3): e380 - e388.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
36/3/286    most recent
afm016v1
Right arrow E-letters: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when E-letters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Holsinger, T.
Right arrow Articles by Plassman, B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Holsinger, T.
Right arrow Articles by Plassman, B.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?